Research

My work concerns the path from an existing structure to a decision about it: how the structure is modelled, what measurements reveal about it, how reliable the resulting prediction is, and when the evidence is sufficient to act. It connects nonlinear computational mechanics, inverse problems, structural reliability and scientific machine learning, and it is meant to remain practical for real structures.

Existing structures, such as bridges, dams and heritage buildings, are assessed, not designed anew, and part of the evidence comes from measurements. This raises questions of identifiability and information: which observations add knowledge, what remains unidentifiable however long we measure, and how much uncertainty survives once the data are in. My work on sensor informativeness and identifiability (2026) examines these questions on a full-scale research bridge. Monitoring and digital twins carry the same questions into continuous assessment.

Predictions then have to support safety decisions. Classical safety formats assume a well-behaved limit state. In nonlinear systems, the failure domain is curved and often non-convex, several failure modes compete, and a factor applied to a material property no longer corresponds to a defined probability. My general partial safety factor theory (2023) extends the classical format to such systems through a notion of homogeneity, so that engineers can keep a format they already know how to apply.

All of this rests on the mechanics: nonlinear finite element and finite–discrete element formulations, constitutive and fracture models, and the numerical methods needed to follow a structure beyond its peak load. I use demanding systems from real infrastructure as benchmarks, because they show where algorithms fail.

Scientific machine learning and high-performance computing make these methods usable at scale: surrogate models and learned material laws that stay physically admissible, and computations fast enough for reliability analysis and monitoring.

The motivation is practical: scientific insight that holds up on a real structure and becomes a method that can be used in an engineering office, in an assessment or in a design code. Work at the university, in engineering practice and in standardization has made this possible, because each setting tests a method in a different way.

Earlier in my career, many of these methods were developed through work on masonry and heritage structures and carried into European standardization. These remain important fields of application within a broader research agenda.