full-field material calibration using ls-opt

21
Full-Field Material Calibration Using LS-OPT ยฎ Nielen Stander, Anirban Basudhar, Imtiaz Gandikota, Suri Bala Sophie Du Bois, Denis Kirpicev H Keshtkar, A Patil, A Sheshadri, P Du Bois (Fiat Chrysler Automobiles) German LS-DYNA Forum Bamberg, Germany October 16, 2018

Upload: others

Post on 12-Apr-2022

6 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Full-Field Material Calibration Using LS-OPT

Full-Field Material Calibration Using LS-OPTยฎ

Nielen Stander, Anirban Basudhar, Imtiaz Gandikota, Suri Bala Sophie Du Bois, Denis Kirpicev

H Keshtkar, A Patil, A Sheshadri, P Du Bois (Fiat Chrysler Automobiles)

German LS-DYNA Forum Bamberg, Germany

October 16, 2018

Page 2: Full-Field Material Calibration Using LS-OPT

Parameter Identification: Overview

โ€ข New curve matching algorithm

Dynamic Time Warping

โ€ข Digital Image Correlation

Nearest Neighbor Cluster: Reduce resources

โ€ข Post-processing

Automated Contour History display (LS-PrePost) using Similarity Measure

Page 3: Full-Field Material Calibration Using LS-OPT

Material Calibration: Introduction

๐‘ง๐‘ง

๐œ‘๐œ‘(๐’™๐’™), ๐œ‘๐œ‘๏ฟฝ

Computed curve: ๐œ‘๐œ‘(๐’™๐’™, ๐‘ก๐‘ก)

Response Surface constructed for each interpolated matching point

Test results

Test curve ๐œ‘๐œ‘๏ฟฝ(๐‘ก๐‘ก)

Material parameters ๐’™๐’™

Prof. S. Kumar, Dr. H. Ghassemi-Armaki (Brown U.) Prof. F. Pourboghrat (OSU)

Experiment (single crystal micropillar)

Result FE model

Test

min๐’™๐’™

๏ฟฝ(๐œ‘๐œ‘๐‘—๐‘— ๐’™๐’™ โˆ’ ๐œ‘๐œ‘๏ฟฝ๐‘—๐‘—)2๐‘›๐‘›

๐‘—๐‘—=1

Simulation

๐’™๐’™โˆ—

Euclidean Distance

Page 4: Full-Field Material Calibration Using LS-OPT

Calibration: Computational challenges

Noise Failure model: GISSMO โ€• element erosion a discrete process

Hysteresis Material 125 โ€• Loading/Unloading (5 cases)

Partial Matching Failure model: GISSMO โ€• post-failure oscillation of coupon

Experimental and computational results can be difficult to compare

Page 5: Full-Field Material Calibration Using LS-OPT

โ€ข DTW calculates the distance between two data sets, which may vary in time, via its corresponding warping path.

โ€ข This path is the result of the minimum accumulated distance which is necessary to traverse all points in the curves.

โ€ข The matching is end-to-end.

โ€ข While the Euclidean distance measure is a strict one-to-one mapping, DTW also allows one-to-many mappings.

โ€ข Mathematically, optimize the path:

๐ท๐ท๐ท๐ท๐ท๐ท(๐‘ƒ๐‘ƒ, ๐‘„๐‘„) =1๐‘™๐‘™

๐‘š๐‘š๐‘š๐‘š๐‘š๐‘š๐‘Š๐‘Š

๏ฟฝ ๐›ฟ๐›ฟ(๐‘ค๐‘ค๐‘–๐‘–)๐‘™๐‘™

๐‘–๐‘–=1

Sophie du Bois, Denis Kirpicev 2018

Addressing noise: Dynamic Time Warping

Page 6: Full-Field Material Calibration Using LS-OPT

Dynamic Time Warping: DTW mapping

Contour Mapping

โ€ข Multi-point histories: Apply to multiple points (full field): ๐œบ๐œบ vs. force

โ€ข Use DTW map to construct

test contours for comparison simulation experiment

Best fit DTW map

Poor fit DTW map

Sophie du Bois, Denis Kirpicev

Simulated GISSMO model: force-displacement curves for tensile test

DTW

dis

tanc

e

Page 7: Full-Field Material Calibration Using LS-OPT

Dynamic Time Warping: Partial curves

โ€ข In DTW, red connectors are summed

โ€ข Curve length difference artificially distorts mismatch

โ€ข Truncation required

Partial curve pairs can distort the DTW result

Page 8: Full-Field Material Calibration Using LS-OPT

Example: GISSMO model

The GISSMO failure model requires special treatment for curve matching

Itera

tions

โ€ข Parameters: 7, Material Model: GISSMO โ€“ Uses discrete (element-by-element)

erosion

โ€ข Curve Matching โ€“ Dynamic Time Warping (DTW) โ€“ Does not address partial curves โ‡จ

Truncate Force history at failure

โ€ข Optimization โ€“ SRSM (fast local optimizer)

Shear: single case calibration history

Page 9: Full-Field Material Calibration Using LS-OPT

Calibration: GISSMO model

Tensile Notch

Shear Punch

In industry, the calibration of the GISSMO model typically involves multiple cases

Optimal match

Courtesy: FCA

Page 10: Full-Field Material Calibration Using LS-OPT

Digital Image Correlation

Page 11: Full-Field Material Calibration Using LS-OPT

Full field test result (4557 pts) from optical scan is mapped and tracked

DIC data: deformation states

Local deformation

๐‘ก๐‘ก = 0

Digital Image Correlation (DIC)

Alignment

Align and map optical data to the Finite Element model

Page 12: Full-Field Material Calibration Using LS-OPT

Digital Image Correlation: LS-OPT technologies (1)

โ€ข Alignment in 3D of test to FE model. Least Squares solution:

๐‘ฟ๐‘ฟ1:Test pts (subset), ๐‘ฟ๐‘ฟ2: FE model pts, ๐‘ป๐‘ป: transform, ๏ฟฝฬ‚๏ฟฝ๐‘ : Isotropic scaling. Typically 3 or 4 points

โ€“ Alternative: LS-PrePostยฎ to translate, rotate and scale test points.

โ€ข Map: Test โ†’ FE mesh: โ€“ Exact Nearest Neighbor (bin tree) search and

element interpolation (107 โ†’ 107 pts). (Practice: ~ 106)

โ€ข Optimization: Minimize Similarity Measure:

min๐‘ฅ๐‘ฅ

๏ฟฝ ๐ท๐ท๐ท๐ท๐ท๐ท(๐‘ƒ๐‘ƒ, ๐‘„๐‘„)๐‘๐‘

๐‘๐‘๐‘๐‘๐‘–๐‘–๐‘›๐‘›๐‘๐‘๐‘๐‘

๐‘๐‘=1

min๐‘ป๐‘ป,๏ฟฝฬ‚๏ฟฝ๐‘

๏ฟฝฬ‚๏ฟฝ๐‘ ๐‘ฟ๐‘ฟ1๐‘ป๐‘ป โˆ’ ๐‘ฟ๐‘ฟ2

Align Test points

Page 13: Full-Field Material Calibration Using LS-OPT

Validation of a Synthetic Problem

๐’„๐’„โˆ— ๐’๐’โˆ—

Start 4.0 0.9

PCM 0.502 0.501

DF 0.500 0.500

DTW 0.497 0.499

Exact 0.500 0.500

๐œ€๐œ€๐‘ฅ๐‘ฅ๐‘ฅ๐‘ฅ ๐œ€๐œ€๐‘ฆ๐‘ฆ๐‘ฆ๐‘ฆ Forc

e

Test pts & nodes coincide

Material 24 with Hockett-Sherby flow curve extrapolation

๐‘“๐‘“ ๐œ€๐œ€๐‘๐‘ = ๐ด๐ด โˆ’ ๐ต๐ต๐‘’๐‘’โˆ’๐‘๐‘๐œ€๐œ€๐‘๐‘๐‘๐‘๐‘›๐‘›

โ€ข ๐’„๐’„ and ๐’๐’ are variables

Full-field deformation

Page 14: Full-Field Material Calibration Using LS-OPT

Distance vs. parameters

Partial Curve Mapping Discrete Frรฉchet Dynamic Time Warping

Different similarity measures compared

Page 15: Full-Field Material Calibration Using LS-OPT

Example 1: DIC Validation: Punch example Calibrate GISSMO material properties using strains/transverse displacement

DIC data

Courtesy: FCA

Page 16: Full-Field Material Calibration Using LS-OPT

๐‘ข๐‘ข๐‘ง๐‘ง

๐œ€๐œ€1

๐œ€๐œ€2

Example 1: DIC Validation: Punch example

Computed Test Difference

The calibration was done using a Force-Displacement similarity match (GISSMO)

Courtesy: FCA

Page 17: Full-Field Material Calibration Using LS-OPT

Digital Image Correlation: Nearest Neighbor Cluster 4557 points 1063 points

โ€ข Accuracy and cost โ€ข Nearest Neighbor Clustering

โ€“ Pre-processing feature โ€“ Reduce resources for large point set

(~106) โ€ข Storage space โ€ข CPU time: mapping is done at each

time step (vanishing nodes/points) โ€“ Nodal 1-to-1 map โ€“ Can also apply a proximity tolerance for

removing outlier points โ€ข Algorithm (๐’•๐’• = ๐ŸŽ๐ŸŽ)

โ€“ Nearest node to each point โ†’ reduced node set.

โ€“ Prune reduced node set โ†’ nearest points โ€“ 1-to-1 map

Page 18: Full-Field Material Calibration Using LS-OPT

Digital Image Correlation: Nearest Neighbor Cluster

โ€ข Accuracy and cost โ€ข Nearest Neighbor Clustering

โ€“ Pre-processing feature โ€“ Reduce resources for large point set

(~106) โ€ข Storage space โ€ข CPU time: mapping is done at each

time step (vanishing nodes/points) โ€“ Nodal 1-to-1 map โ€“ Can also apply a proximity tolerance for

removing outlier points โ€ข Algorithm (๐’•๐’• = ๐ŸŽ๐ŸŽ)

โ€“ Nearest node to each point โ†’ reduced node set.

โ€“ Prune reduced node set โ†’ nearest points โ€“ 1-to-1 map

๐Ÿ‘๐Ÿ‘. ๐Ÿ–๐Ÿ– ร— ๐Ÿ๐Ÿ๐ŸŽ๐ŸŽ๐Ÿ“๐Ÿ“ points ๐Ÿ•๐Ÿ•๐Ÿ•๐Ÿ•๐Ÿ•๐Ÿ•๐Ÿ“๐Ÿ“ points = ๐Ÿ๐Ÿ. ๐Ÿ—๐Ÿ—๐Ÿ—

Enlarged

DIC Test points (full set)

Nearest test points

Page 19: Full-Field Material Calibration Using LS-OPT

Example 2: Tensile test

Computed

DIC Test points

๐œ€๐œ€๐‘ฆ๐‘ฆ๐‘ฆ๐‘ฆ Nearest test points

DIC

๐œ€๐œ€๐‘ฅ๐‘ฅ๐‘ฅ๐‘ฅ FE model

The contour comparison uses Dynamic Time Warping: ๐Ÿ‘๐Ÿ‘. ๐Ÿ–๐Ÿ– ร— ๐Ÿ๐Ÿ๐ŸŽ๐ŸŽ๐Ÿ“๐Ÿ“ DIC points

โ€ข Reduces 380,000 DIC points to 7275 points with nodal neighbors

โ€ข Reduces extraction time from 2 hours โ†’ 6 minutes

Page 20: Full-Field Material Calibration Using LS-OPT

LS-OPT DIC calibration feature summary (v6.0)

โ€ข DIC Interfaces: โ€“ gom/ARAMIS

โ€ข v6 CSV โ€ข v7 XML

โ€“ Fixed Format (LS-PrePost) โ€“ Free Format (LS-OPT/GenEx parser)

โ€ข LS-DYNA interface โ€“ Multi-point histories (d3plot) โ€“ Entities

โ€ข Nodal โ€ข Shell โ€ข Solid

โ€“ Exact nearest neighbor point mapping (~107 pts). Test pt โ†’ FE pt

โ€ข Curve similarity methods โ€“ Euclidean, Frรฉchet, DTW, PCM

โ€ข Filtering โ€“ Online filtering (SAE, Ave)

โ€ข GUI โ€“ Test pre-view โ€“ Test alignment

โ€“ Strain fringe plot (LS-PrePost) โ€ข Simulation โ€ข Experiment โ€ข Error

DTW

Page 21: Full-Field Material Calibration Using LS-OPT

Outlook

โ€ข General feature: Improved pre-viewing/pre-processing of experimental data.

Interactive filtering and truncation of test results

โ€ข Partial DTW-based curve mapping

DTW-LCS method

โ€ข Further speedup

Multiple similarity responses typically have the same mapping