Subject: ✈ MDO Lab Newsletter

View this email online if it doesn't display correctly
MDO Lab Newsletter #4, Mar 2018

New methods and tools for surrogate modeling
SMT: An open source surrogate modeling toolbox
We have recently released the Surrogate Modeling Toolbox, which was developed in collaboration with ONERA, NASA, and ISAE-SUPAERO. This is a Python-based open source framework that implements various surrogate modeling methods, including Kriging and RBF. The framework focuses on using and providing gradients, and it includes two newly developed methods that scale well for large problems (see the two papers below).
Gradient-enhanced kriging for high-dimensional problems 
Optimized CRM
Surrogate models are known to suffer from the curse of dimensionality. To address this issue, we developed a new surrogate modeling technique (gradient-enhanced kriging with partial least squares: GE-KPLS) that scales much better with the number of inputs with no loss in accuracy. This builds on the KPLS approach and is included in the SMT package.

Surrogate-based airfoil analysis and design
We used the surrogate modeling approach mentioned above to create a surrogate model based on RANS solutions for all the UIUC database airfoils and airfoils generated using a design of experiments. As a result, we can analyze airfoils in both subsonic and transonic regimes in about one hundredth of a second, and optimize airfoil shapes in a few seconds, with an error under 0.8 counts for the subsonic regime and under 2.5 counts for the transonic regime.

                                [ Paper ]
A fast-prediction surrogate model for large datasets

TACS wingbox model
We developed yet another surrogate modeling technique called regularized minimal-energy tensor-product splines (RMTS). Unlike the GE-KPLS mentioned above, this one does not scale well with the number of variables. However, it addresses another issue with surrogate models, which is the scalability with the number of training points. It is also shown to be much more accurate than kriging for a given number of training points. RMTS is specially suitable for construction accurate drag polars, and it is included in the SMT package.

Other recent journal articles
OpenAeroStruct: An open-source tool for aerostructural optimization

TACS wingbox model
OpenAeroStruct is a lightweight Python tool that performs aerostructural optimization of lifting surfaces. It uses a vortex lattice method for the aerodynamics analysis coupled to a spatial beam model for the structural analysis. The optimization is efficient thanks to a coupled adjoint method for computing the gradients. OpenAeroStruct is ideal for those who want to get started in wing design, OpenMDAO, or want to benchmark MDO methods.

                                                  [ Documentation ]    [ GitHub repository ]
Benchmark of optimization algorithms in wing aerodynamic design
In this paper, we benchmark both gradient-based and gradient-free optimization algorithms for computational fluid dynamics based aerodynamic shape optimization problems based on the Common Research Model wing geometry. We also investigate the existence of multiple local minima by solving the same optimization problem starting from a series of randomly generated initial geometries, as well as a wing based on the NACA 0012 airfoil with zero twist.

Boundary layer ingestion analysis

TACS wingbox model
As a precursor to a simultaneous optimization of the aerodynamic shape and propulsion system, we studied modeling of the airframe propulsion integration in a fully coupled way using RANS. We previously published a manuscript on propulsion cycle analysis with adjoint derivatives . The final goal is to perform the aeropropulsive optimization of a tailcone thruster configuration. 

You may unsubscribe or change your contact details at any time.