Cross-Covariance-Based Model Reduction P. Benner, S. Grundel, C. Himpe Max Planck Institute Magdeburg Computational Methods in System and Control Theory Group Simulation of Energy Systems Team MS28 – Model Reduction Methods for Simulation and (Optimal) Control ENUMATH – European Conference on Numerical Mathematics 2017–09–26
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Cross-Covariance-Based Model ReductionP. Benner, S. Grundel, C. Himpe
Max Planck Institute MagdeburgComputational Methods in System and Control Theory Group
Simulation of Energy Systems Team
MS28 – Model Reduction Methods for Simulation and (Optimal) ControlENUMATH – European Conference on Numerical Mathematics
Repeated simulation of gas transportation networks.
Transient behaviour on typical temporal resolutions.
Test families of possible supply-demand scenarios.
Uncertainty quantification for unsteady supply.
Short-term dispatch forecasts.1
C. H. Combined State and Parameter Reduction for Nonlinear Systems with an Application in Neuroscience.Westfalische Wilhelms Universitat Munster, 2016.
C. Himpe, [email protected] Cross-Covariance-Based Model Reduction 2/25
Average pressure / mass-flux over pipe cross-section area.
Conservation of momentum / mass
Hyperbolic (coupled transport)
Nonlinear (friction term)
Additional nonlinearities (compressibility)2
S. Grundel, N. Hornung, B. Klaassen, P. Benner and T. Clees. Computing Surrogates for Gas Network Simulation UsingModel Order Reduction. In: Surrogate-Based Modeling and Optimization, Applications in Engineering: 189–212, 2013.
C. Himpe, [email protected] Cross-Covariance-Based Model Reduction 5/25
Spatial Discretization and Index Reduction3,4:(E1 00 1
)(phqh
)=
(0 A1
A2 0
)(phqh
)+
(fp(ph, up)
fq(ph, qh, up)
)+
(0 B1
B2 0
)(upuq
)(ypyq
)=
(C1 00 C2
)(phqh
)Descriptor system,
Index-1 differential-algebraic equation (DAE).
Analytic index reduction to index-0 → implicit ODE.
3S. Grundel, L. Jansen, N. Hornung, T. Clees, C. Tischendorf and P. Benner. Model Order Reduction of Differential
Algebraic Equations Arising from the Simulation of Gas Transport Networks. In: Progress in Differential-Algebraic Equations,Differential Equation Forum: 183–205, 2014.
4S. Grundel, N. Hornung and S. Roggendorf. Numerical Aspects of Model Order Reduction for Gas Transportation
Networks. In: Simulation-Driven Modeling and Optimization: 1–28, 2016.
C. Himpe, [email protected] Cross-Covariance-Based Model Reduction 6/25
U∗(1...n), V(1...n)∗ are left and right principal directions.
Reconstrucing projection: U =(U1 U2
)Galerkin reducing projection: V1 = Uᵀ
1
Petrov-Galerkin reducing projection8: V =(V1 V2
)8
D.C. Sorensen and A.C. Antoulas. The Sylvester equation and approximate balanced reduction. Linear Algebra and itsApplications, 351–352: 671–700, 2002.
C. Himpe, [email protected] Cross-Covariance-Based Model Reduction 17/25
Interfaces for: Solver, inner product kernels & distributed memoryNon-Symmetric option for all cross GramiansCompatible with OCTAVE and MATLABVectorized and parallelizableOpen-source licensedFunctional design
More info: http://gramian.de
C. Himpe, [email protected] Cross-Covariance-Based Model Reduction 21/25
Acknowledgment:Supported by the German Federal Ministry for Economic Affairs andEnergy, in the joint project: “MathEnergy – Mathematical KeyTechnologies for Evolving Energy Grids”, sub-project: Model OrderReduction (Grant number: 0324019B).
C. Himpe, [email protected] Cross-Covariance-Based Model Reduction 25/25