25.11.2025
Teknikum Case Study
Optimizing Mechanical and Electrical Performance of Material Compositions
This case study showcases how our partner, a Finnish manufacturer of advanced polymer and rubber-based materials, made significant strides with the BOMP platform to identify optimal material compositions that enhance mechanical strength and electrical conductivity for their multifaceted applications. Our partner’s team aimed to improve the overall performance of a specific compound formulation while adhering to tight manufacturing constraints. The goal was to optimize hardness, electrical conductivity, and tensile strength within a constrained design space and while excluding certain legacy raw materials.
Optimizing multiple target properties simultaneously, especially under strict compositional limits, presents a major challenge for traditional R&D workflows. Manual analysis of the high-dimensional dataset and experimental iteration would have been time-consuming, resource-intensive, and prone to missing non-obvious combinations that yield superior performance.
Using the BOMP algorithm, the dataset was analyzed to uncover the relationships between the input compositions and the key performance targets. The algorithm trained a machine learning model on our partner's experimental data and explored the design space to suggest new compositions predicted to outperform existing material compositions. The BOMP optimization process automatically adhered to our partner's processes and constraints, generating a curated list of new candidate formulations. Each was predicted to exceed the company’s performance thresholds for critical properties.
Within a few seconds of running time, BOMP was able to identify, and suggest a ranked list of high-performing candidate compositions, a detailed breakdown of the influence of each component as well as predictive metrics showing how suggested formulations compared to existing data. BOMP’s predictions revealed clear patterns within the dataset. The analysis confirmed which materials play a dominant role in boosting material performance, aligning with our partner's empirical observations.
More importantly, BOMP did not yield just a single “best” formulation. Instead, it produced a cluster of high-performing options, giving the partner flexibility to select the optimal formulation based on cost, manufacturability, or supply-chain considerations. By using BOMP, our partner successfully accelerated its materials discovery process through significantly reducing R&D time by eliminating the need for extensive manual testing. They identified new, high-performance formulations meeting strict electrical and mechanical targets while validating existing hypotheses about key component behavior with quantitative evidence. Finally, they were able to enhance their decision-making through data visualization and predictive modeling.
The Proof-of-Concept demonstrated that the BOMP optimization can deliver substantial value even with a limited dataset, empowering partners to innovate faster and with greater confidence.
Optimizing multiple target properties simultaneously, especially under strict compositional limits, presents a major challenge for traditional R&D workflows. Manual analysis of the high-dimensional dataset and experimental iteration would have been time-consuming, resource-intensive, and prone to missing non-obvious combinations that yield superior performance.
Using the BOMP algorithm, the dataset was analyzed to uncover the relationships between the input compositions and the key performance targets. The algorithm trained a machine learning model on our partner's experimental data and explored the design space to suggest new compositions predicted to outperform existing material compositions. The BOMP optimization process automatically adhered to our partner's processes and constraints, generating a curated list of new candidate formulations. Each was predicted to exceed the company’s performance thresholds for critical properties.
Within a few seconds of running time, BOMP was able to identify, and suggest a ranked list of high-performing candidate compositions, a detailed breakdown of the influence of each component as well as predictive metrics showing how suggested formulations compared to existing data. BOMP’s predictions revealed clear patterns within the dataset. The analysis confirmed which materials play a dominant role in boosting material performance, aligning with our partner's empirical observations.
More importantly, BOMP did not yield just a single “best” formulation. Instead, it produced a cluster of high-performing options, giving the partner flexibility to select the optimal formulation based on cost, manufacturability, or supply-chain considerations. By using BOMP, our partner successfully accelerated its materials discovery process through significantly reducing R&D time by eliminating the need for extensive manual testing. They identified new, high-performance formulations meeting strict electrical and mechanical targets while validating existing hypotheses about key component behavior with quantitative evidence. Finally, they were able to enhance their decision-making through data visualization and predictive modeling.
The Proof-of-Concept demonstrated that the BOMP optimization can deliver substantial value even with a limited dataset, empowering partners to innovate faster and with greater confidence.
