University of Liverpool (UK)
Sebastian Kruse was born in Berlin/Germany and graduated with Master of Science in mathematics
from Leibniz University Hanover, where he also got his bachelor’s degree. Both his theses dealt with
the theoretical analysis of iterative solution methods for particular nonlinear optimization problems
under specific constraints.
During his studies he worked as a research assistant, inter alia, in the institute for risk and reliability
under the leadership of Prof. Dr. Michael Beer. Further, he spent two years working at MTU
Maintenance GmbH. As student employee in the major project ‘sfc-optimized Compressor
Maintenance’, a cooperation between MTU and the Technical University Braunschweig, he was part
of the organizational unit for ‘Industrial Engineering’, where his assignment was to develop an
algorithm for performance-optimized assembly of high-pressure compressors with technical and
legal restrictions. These experiences offered him the opportunity to gain deeper insight into various
engineering disciplines and sparked his interest in the associated problems and questions.
Now located at the University of Liverpool and primarily supervised by Dr. Edoardo Patelli within the
context of the DyVirt project, Sebastian’s particular research project aims at the development of
complex load models to capture spatial and temporal variations of loads on large, complex
structures in dynamic environments. Within his research he is particularly interested in the (wave,
sea current and) wind loads on offshore wind turbines. At the core of the envisaged approach lies
Bayesian model updating in order to combine fragmentary data and available expert knowledge into
an appropriate load model. A method to find the best possible point estimate for the Evolutionary
Power Spectra representing these complex loading scenarios is also to be developed. Currently
Sebastian is examining whether and how the technique of compressive sensing to achieve a low
dimensionality of the problem and the use of copulas for modelling correlations are suitable in this
context.