The optimal design of mechanical structures is hampered by the imperfect knowledge of the environmental conditions under which they will operate. The systematic incorporation of the resulting uncertainties currently relies on the intensive use of complex simulators.

Under this constraint, IFPEN is designing wind turbine anchoring technologies and floaters that reconcile:

  • operating robustness,
  • and cost with a view to large-scale production.

Since simulation based optimization can be costly in terms of calculation time, researchers are working on a more “economic” exploration of possibilities, particularly concerning environmental conditions. Moreover, the relative scarcity of failures implies that the simulated cases need to be sufficiently numerous to expect the observation of such phenomena and to adapt the design of the structure accordingly.

These two challenges were tackled respectively:

  • using dimension reduction techniques, for random processes modeling environmental conditions,
  • and using accelerated Monte-Carlo simulation methods, to reduce the number of scenarios required to estimate the probability of failure(1).

In addition, since the optimization constraints involve the probability of failure, and the resolution method employed requires the calculation of their derivative, further work was conducted in this respect(2).

Lastly, still with the aim of reducing calculation times, an alternative was applied to the design of an offshore wind turbine anchoring solution, resistant to fatigue: using probability constraint approximations based on the theory of extreme values.

Thereby, through the provision and introduction of new methodologies, but also computing equipment and algorithms, IFPEN is helping to improve design tools for floating wind turbine supports, working closely with industrial players in the sector.


Éolien flottant
Floating wind turbine

(1)  A. Murangira, M. Munoz Zuniga, T. Perdrizet. Structural reliability assessment through metamodel based importance sampling with dimension reduction.
>> https://hal.archives-ouvertes.fr/hal-01230454/document, 2015.
(2)  W. van Ackooij, I. Aleksovska, M. Munoz Zuniga. Submitted to Set-Valued and Variational Analysis, 2017.


Scientific contact: Miguel Munoz Zuniga




Research engineer in uncertainty quantification and optimization
PhD in applied mathematics