Challenges in realizing 3rd generation multidisciplinary design optimization

Figure 1.  Project overview: Modeling skin heat exchanger, adapted from source [22]

Figure 2.  Evolution of the MDO systems, source [11]

Figure 3.  Coupling external solvers/codes to KRATOS using a detached interface. Source[GitHub CoSimulationApplication]

Figure 4.  Data exchange types: a) distributed b) centralized data exchange architectures

Figure 5.  Different scenarios for running CoSimulation in distributed environments (4 MPI-processors), derived from [8]

Figure 6.  Analytical Example

Figure 7.  2D NACA0012 on spring

Figure 8.  3D Onera M6 wing on spring

Figure 9.  3D flexible Onera M6 wing on spring

Figure 10.  Surrogate-based optimization workflow using OptApp

Figure 11.  The CFD-mesh around the airfoil

Figure 12.  The FSI problem under consideration

Figure 13.  The initial airfoil shape obtained by fitting a NACA 0012 profile, the predicted optimal shape using the initial DoE, and the optimization result after running MDO-TS

Figure 14.  Streamlining MDO problem formulation with a knowledge graph

Figure 15.  Objects and properties from MDO data are associated with Wikidata entries to increase FAIRness

Figure 16.  Knowledge graph extracted from the analytical example. The root node $\texttt{structural model}$ represents a structural model, and its input parameters, such as $\texttt{chord}$ and $\texttt{lift}$, are connected with the model by the $\texttt{has input}$ relation. In turn, $\texttt{chord}$ and $\texttt{lift}$ are instances of the class $\texttt{Variable}$, which is depicted with orange arrows