黑料福利网

Soft materials are everywhere in daily life, yet they behave in surprisingly complex ways. Some flow like liquids, others stretch like elastic solids, and some resist movement entirely until enough force is applied. This unpredictable behavior makes it hard to model what happens during real processes such as mixing, squeezing, or coating. Engineers and scientists depend on these predictions to design efficient production methods, but traditional approaches often struggle to capture this complex behavior.

The hidden problem: uncertainty everywhere

Even the best experiments and models are never perfect. Measurements vary, and models must simplify reality to remain usable. Historically, researchers tried to minimize these imperfections as much as possible. This thesis of Aricia Rinkens takes a different stance: uncertainty is unavoidable, so instead of ignoring it, it should be used as part of the analysis. Treating uncertainty as useful information leads to more realistic and trustworthy predictions.

A Bayesian shift: from single answers to ranges

At the heart of this research lies Bayesian uncertainty quantification, a probabilistic framework that updates predictions as new data becomes available. Instead of producing a single 鈥渂est guess,鈥� this approach provides a range of likely outcomes and shows how confident we can be in those predictions. It also allows different material models to be compared in a fair way, naturally favoring models that explain the data well without being unnecessarily complicated.

Two everyday experiments, big insights

To test the approach, the research focuses on two flow scenarios that are both simple and industrially relevant. The first is squeeze flow, where a material is compressed between two plates, similar to processes such as rolling dough or compressing gels. The second is the pendant drop experiment, where a droplet hangs from a needle, a method commonly used in industry to study fluids used in printing and coating. The research shows that even with relatively simple experimental setups, reliable predictions are possible, as long as the uncertainties are properly quantified and included in the analysis.

Smarter models and practical use

Rinkens also evaluates commonly used MCMC sampling methods and establishes practical criteria to ensure reliable results without excessive computational cost. In addition, Bayesian model selection is applied to complex flow scenarios, explicitly accounting for both measurement noise and model error, which improves predictive performance. By applying the framework to an industrially relevant case, pendant drop tensiometry, the research clearly demonstrates that the method is not just theoretical, but practically applicable.

A new mindset for soft matter engineering

This thesis shows that embracing uncertainty leads to more realistic predictions, better model selection, and greater practical usability in industry. The main message is clear: uncertainty is not something to eliminate, but something to understand and use. This shift in mindset provides a more reliable foundation for designing, optimizing, and scaling industrial processes involving soft materials, contributing to more efficient and innovative manufacturing.

Title of PhD thesis: . Supervisors: Dr. Nick Jaensson, Dr. Clemens Verhoosel, and Prof. Patrick Anderson.