1 / 15

Lecture 25 Structural Approximation (Fast Reanalysis)

Lecture 25 Structural Approximation (Fast Reanalysis). EGM6365 Structural Optimization 03/12 Given by Shu Shang. Introduction.

lan
Download Presentation

Lecture 25 Structural Approximation (Fast Reanalysis)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Lecture 25Structural Approximation (Fast Reanalysis) EGM6365 Structural Optimization 03/12 Given by Shu Shang

  2. Introduction • In static response,designperturbationfrom x0to x0+ Δxcan causechangeinresponsefromu0to u0+ Δuaswellas stiffness matrix from K0to K0+ ΔK • When the perturbation size Δx is small, we expect Δu and ΔK will also be small • Then, instead of rebuilding stiffness, K0+ ΔK, with a small perturbation, it is possible to approximate Δu with a reasonable accuracy. (Structural approximation) • Since K0is factorized already, we can use the factored matrix K0to approximate Δu. (Fast reanalysis)

  3. Linear Static Response • Linear static response at the initial design x0 • At the perturbed design, • Subtracting the equilibrium equation of the initial design (1) • We approximate Δu into Δ u1by ignoring H.O.T. ΔKΔu (2) • Accurate when Δxis small • This process is fast because K0is already factorized

  4. Improvement of Approximation • Subtracting (2) from (1) • ApproximateΔu2= Δu– Δu1and ignoring H.O.T. ΔK(Δu–Δu1) • If we repeat this process continuously Where the terms Δuiare obtained through the iterative process of solving

  5. Approximation with Scaling • Kirsch and Taye introduced scaling and redistribution • Choose s to minimize ΔKsso that sK0is close to K0+ΔK • Calculate s to minimize the square sum of elements of ΔKs • Then, consider initial design to be sK0 instead of K0, and • Only consider the case where

  6. Total change Δu predicted by this approach

  7. Example: 1D Bar P • Design parameter: cross section area A • Initial design x0=1 • Perturbation Δx=0.25 • Stiffness • Tip displacement • Approximation L

  8. One more iteration • Another approach

  9. Eigenvalue Problem • Vibration or buckling response • At perturbed design • Subtract (3) from (4) and ignore H.O.T. • Pre-multiplying by u0T • Or pre-multiply (4) by and neglect some higher order terms

  10. Example: Mass-spring system • Estimate the effect on the lowest frequency caused by doubling the left mass • Stiffness matrix: • Mass matrix: • The lowest eigenvalue and corresponding eigenvector are • Perturbation: • Exact result:

  11. First approach • Another approach

  12. Problem 1 • Estimate the effect on the lowest frequency caused by an 50% increase in the stiffness of the left spring P

  13. Exact Reanalysis • Calculate Δuexactly (no approximation) • Still it needs to be fast • Let K be the original stiffness matrix, and we have its inverse • Here ΔKis rank one matrix and can be written as uvT • Sherman-Morrison formula • Popular for truss structures, since the change of one truss element leads to a rank-one modification of K

  14. Mehmet A. Akgun,John H. Garcelonand Raphael T. Haftka, Fast exact linear and non-linear structural reanalysis and the Sherman-Morrison-Woodbury formulas, International journal for numerical methods in engineering • Review of the re-invention of the Sherman-Morrison method by different authors over the years from 1950 to 2000

  15. Problem 2 • Solve problem 1 using Sherman-Morrison formula

More Related