A brief introduction to the iHIT Solution Engine.
First, the engine requires an intense discussion of your chemical challenge. Our Expert translates your chemistry into an experimental plan.
The ML algorithm generates a design plan for the experiments. It is your decision to perform the computer-suggested experiments in your laboratories or on our automated, high-throughput formulation system.
Measured data relate product properties to formulation recipes and are collected by the Machine Learning algorithm. The algorithm starts to build a model of your formulation and evaluates its quality by preset criteria. With only few experimental data, the model is incomplete and the algorithm suggests more experiments. This iterative process quickly converges to an optimized solution.
At this point, the algorithm stops the engine. The best formulation is presented that is your formulation for production.
The iHIT Solution Engine supports your product development in a unique way.
A brief analysis of the technology readiness of your product shows that great research ideas often face the valley of death before a product is ready for market entry.
The iHIT Solution Engine optimizes your promising formulations to a market-ready level – fast and highly efficient.
The iHIT Solution Engine is applied to an application of our D-NL-HIT partner DROST Coatings. The challenge was to transform an existing product into a bio-based version.
Expert discussion identified the binder as the sole raw material to be replaced. The information of the existing product along with bio-based binder options is entered into the ML algorithm, which suggests a first set of 10 experiments.
These along with characterization were done at the DROST laboratory in order to assure compatibility to their standard procedures. The algorithm advised 4 iterations and approximately 50 experiments until it concluded that the bio-based binder B-3 adequately replaces the original binder B-0.
It adjusted the recipe for the new binder, too, assuring that the performance of the bio-based product is identical to the original.
Furthermore, the model recommends modifications of the recipe for best matching specific properties such as hiding power, gloss, or water resistance.
The developed model is also a tool to further improve and adjust your formulation with regard to different applications. It grows with more experiments over time.