COMPASS From Data To Understanding

1
COMPASS From Data To Understanding

• Anthony R. Ingraffea Gerd Heber
• Cornell Fracture Group
• SIPS 10th Quarterly
• St. Augustine, FL
• January 31, 2007

2
Outline

• Goals
• Infrastructure
• A quick tour of DDSim Level 1
• Polycrystals
• From data to understanding
• Work in progress

3
Goals

• One-stop shopping for multiscale simulation of
  fatigue crack incubation, nucleation, and
  propagation

• Hosting of underlying characterization data and
  information products (e.g., response surfaces)
• Part of the SIPS software system

4
Infrastructure

• Hosted _at_ Cornell Theory Center
• Hardware
• 1 web server, 2 database servers
• 2x dual-core AMD Opteron, 8 GB RAM
• 24 TB of disk (70 in use)
• ADMM cluster
• 170 Dell Dual Pentium Xeon _at_ 3.6 GHz blades (
  340 processors)
• Software
• Windows Server 2003 x64 Edition
• ASP.NET 2.0
• SQL Server 2005
• JavaDX
• Google Maps

5
CUs Black Box
Low fidelity life prediction

• Low fidelity life predictor
• Level I for Ntotal NMSCNMLC
• Level I for NMSC

Stochastic input
predictor
corrector
High fidelity life prediction

• Hot-spot (HS) iterator
• Level II for NMLC
• Combine conditional probabilities for Ntotal

Low fidelity prior conditional life pdf

• Bayesian Estimator
• Multiscale Level III for NMSC

High fidelity post conditional life pdf
6
6
7
7
8
8
9
9
10
10
11
11
12
(No Transcript)
13
13
14
Features
15
Polycrystal Scale Modeling
Generate first-phase realization
Mesh mstructure w/o particles
Sample texture distribution
BCs for mstructure
Compute strain fields w/o particles cracks
Sample particle distribution
Output FEM
Filter particles for realistic mstructure model
Mesh mstructure w/ particles
Insert particles of interest
16

• Polycrystals
• Geometry Templates
• Particle insertion
• Custom geometries
• Meshing
• FEA
• Visualization
• Analysis

16
17
From Data To Understanding

• Data Sources
• Experiment
• CMUs texture and particle characterization
• Monitoring
• Simulation
• CUs incubation/nucleation databases
• Testing
• NGCs imagery and mechanical testing data
• Data Source Explorer ???
• Google Maps
• GIS

• Information
• Calibration info for
• Crack nucleation meta-models (MSU)
• Multisite FASTRAN (ALCOA)
• Reasoning and prediction models (NGC)
• Services
• Analysis
• Visualization
• Web services
• Understanding
• VV of FCC plasticity model (RPI)
• VV of fatigue mechanisms (CU)

18

• CMU Particle Data
• Example
• 4 samples
• 3 sections / sample
• 2 images / section

18
19

• NGC Imagery
• Test Data
• Image library
• Metadata
• Annotation
• Search
• Analysis tools

19
20
Work in Progress

• Describe the information products (50 complete)
• Define the system scope (75 complete)
• Create a data design (80 complete)
• Choose a logical data model (80 complete)
• Determine system requirements (60 complete)
• Implementation (60) complete
• Integration w/ SIPS system (30)