Artificial Intelligence versus Design of Experiments in Rubber Compounding

Date: December 9, 2021 2:00 pm (ET)


  • Hans-Joachim Graf
    H-JG Consulting

Development of Compounds in Rubber Industry is targeting new raw material usage, adaption to changes in processing or quality and cost performance improvements for example. Very often – if not always – this are conflicting targets. In any case the developer deals with a multifactor optimization problem. Nowadays the discussion is on, to use modern computer techniques like Artificial Intelligence [AI] in compound development. It is thought, that this technique may replace other techniques like Statistic Experimental Design [DoE].

AI handles existing data from any database created in the past. It works with objective functions that are to be minimized or maximized. The answer is a set of solutions, which are defining the best tradeoff between competing objectives, but concluding out historic data.

If there are no data existing for example about a newly developed raw material, DoE is the procedure to answer, whether it can be used or not. This procedure is based on factor – response correlations and regression analysis. An optimum solution or best compromise is the result.

These techniques replacing completely trial and error or one step at a time procedure. AI and DoE have its advantages and disadvantages, which will be discussed in this presentation. In modern rubber development both are needed.