Möglichkeiten und Grenzen

5. GMTTB Jahrestagung 2015 Menschmodellierung zur Verletzungsvorhersage – Möglichkeiten und Grenzen Rainer Hoffmann, carhs gmbh www.carhs.de
Ist der Mensch berechenbar?
Der Mensch ist berechenbar!
School Bus Accident Simulation using CAL3D
Rainer Hoffmann, ME 710, Wayne State University, 1984
5th
dedicated sessions
Inhalt
Crash Vic6m Simula6on Status der Modelle Möglichkeiten und Grenzen der Verletzungsvorhersage ZukünDige Entwicklungen
Crash VicUm SimulaUon
1973: 1-­‐D Lumped Mass Model: Lobdell Thorax Model
1985: 3-­‐D Rigid Body Model: MADYMO3D
2013: 3-­‐D Finite Element Model: Hybrid III 50th%ile male
:sledoM
elameF dna elaM tlud
nairtsedeP dna tnapucc
seziS egraL dna elddiM ,llam
Crash VicUm SimulaUon
Dummy Modelle
05MA
50FA
aliava( sledoM nairtsedeP
Mensch Modelle
50FA
05MA
59MA
)elbaliava( sledoM tnapuccO
Anwendungsgebiete Mensch Modelle
Innova6ve Rückhaltesysteme reale Menschen (außerhalb 5th, 50th, 95th) Integrale Sicherheit Dummy-­‐Entwicklung Kopfverletzungen Unfallrekonstruk6on
Mercedes-Benz S-Class
Seat Cushion Airbag
• Seat Cushion Airbag reduces risk for submarining in
reclined excecutive rear seats
• Activated by crashsignal when excecutive rear seats are reclined
17 Safety Features of the new S-Class | 21.05.2014
Mercedes-Benz
Quelle: Bogenrieder, SafetyUpDate 2014
Inhalt
Crash Vic6m Simula6on Status der Modelle Möglichkeiten und Grenzen der Verletzungsvorhersage ZukünDige Entwicklungen
THUMS (Total Human Model for Safety) 1. Model Overview
entwickelt von Toyota seit 2000 Base Models:
aktuell: Version 4 (1.8 Mio. Elements)
• Adult Male and Female
Tissue Damage and Rupture: • Occupant and Pedestrian
• Internal Organ Injury (Contusion, Lacera6on) ... Strain Based Indicator • Small, Middle and Large Sizes
• Brain Injury (Diffuse Axonal Injury) ... Strain Based Indicator •
Skeletal Injury (Bony Fracture) ... Element Elimina6on
AM95
AM50
AF05
Occupant Models (available)
AF05
AM50
AM95
Pedestrian Models (available)
ifiedGHBM Occupant(Global Human Body Model)
Models
entwickelt vom GHBMC seit 2006 aktuell: M50 4.1.1
Detail Considered
More
~2.0M deformable
elements
Detailed GHBMC M50 Occupant
GHBM (Global Human Body Model) HUMAN MODELING AND SIMULATION IN AUTOMOTIVE ENGINEERING
Vorhersage von Crash Induced Injuries (CII) Crash-Induced
Injury (CII) Prediction Capability
8
Stand der Vorhersagefähigkeit dokumen6ert
To assess the injury prediction capability of GHBMC models, we have established 7
capability levels as defined below:
Level
0
1
2
3
4
Capability Subcategories
Model detail sufficient, test data available, injury mechanism
understood, correlation carried out
Model detail sufficient, test data available, injury mechanism
understood, but validation work is incomplete or inconclusive
Model detail sufficient, but test data unavailable or insufficient
Model detail insufficient, test data available, additional
modeling should help predict this CII
Model detail insufficient, test data unavailable; additional
modeling effort and test data should help predict this CII
5
Injury mechanism needs some more investigation
6
Injury mechanism needs extensive additional investigation
© 2014 carhs.training gmbh
8
GHBM (Global Human Body Model) HUMAN MODELING AND SIMULATION IN AUTOMOTIVE ENGINEERING
M50Beispiel: CII Capability: Level „0“
Model v4.1 – CII Capability: Level “0”
9
Crash-Induced Injury (CII) Description
GHBMC
M50
Capability
Body Region
Main
Head
Head
Neck
Neck
Thorax
Abdomen
Abdomen
Plex
Plex
Plex
Plex
Plex
Plex
Plex
Plex
Sub
Skull Fracture
Facial Bone Fracture
Intervertebral Disc
Ligament Injury
Rib Cage Injuries
Solid Organ Injury
Solid Organ Injury
Pelvis
Pelvis
Thigh, Knee, Leg
Thigh, Knee, Leg
Thigh, Knee, Leg
Foot
Foot
Foot
Cortical Layer, Diploe Layer, Vault, Base
Disc Injury
Rib Fracture
Liver Injury
Spleen Injury
Pelvis, pubic rami fracture
Pelvis, hip fracture
Proximal femur fracture
Mid-shaft femur fracture
Distal femur fracture
Calcaneus fracture
Talus fracture
Ankle and sub-talar joint injury
© 2014 carhs.training gmbh
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Quelle:
9 J.T. Wang, SAE G&I 2014
GHBM CII Coverage of GHBMC M50 Model based on 1998-2009
(Global Human Body Model)
NASS-CDS Data
GHBMC M50 model is validated for • GHBMC M50 model is validated for
38%• of38% of all crash-­‐induced AIS2+ injuries all crash-induced AIS2+ injuries
13%• AIS5+
injuries
13% AIS5+ injuries • GHBMC
M50 model is detailed enough
GHBMC M50 model is detailed enough for simula6ng for simulating
Model
based
on 1998-2009
• 80% of all crash-­‐induced AIS2+ injuries 80% of all crash-induced AIS2+ injuries
63% AIS5+ injuries 63%• AIS5+
injuries
• Phase II targets CII capability equal or
better than M50 Model
ries• Should additional funding become
available in the future, we will
h
Conduct more validation work for Level
1&2 CIIs
ries
Increase the detail level of the model to
predict Level 3&4 CIIs
Quelle: J.T. Wang, SAE G&I 2014
SAE INTERNATIONAL
SUFEHM (Strasbourg University FE Head Model)
SUFEHM PRESENTATION
50th%ile Adult Human Head, 13.208 Elemente Brain
Membranes
(Viscoelastic G0=49kPa, G =16.7kPa, β=145s-1)
(Elastic E=31.5MPa, =0.23)
CSF
(Elastic E=12kPa, =0.49)
Brainstem
(Viscoelastic G0=49kPa, G =16.7kPa,
β=145s-1)
Skull
(Shell elements, composite
law with failure criterion)
Scalp
(Elastic E=16.7MPa, =0.42)
Face
(rigid)
Quelle: R. Willinger, carhs Seminar T141
Inhalt
Crash Vic6m Simula6on Status der Modelle Möglichkeiten und Grenzen der Verletzungsvorhersage ZukünDige Entwicklungen
Garbage
in,
garbage
out.
„Garbage In-­‐garbage Out“ Paradigma
GARBAGE DATA
PERFECT MODEL
GARBAGE RESULTS
PERFECT DATA
GARBAGE MODEL
GARBAGE RESULTS
EVERYTHING SHOULD BE MADE AS SIMPLE AS POSSIBLE, BUT NOT SIMPLER. ALBERT EINSTEIN
How simple is simple enough?
Beispiel: Hirnverletzung
1 Element -­‐ Head Injury Criterion (HIC) 10K -­‐ 300K Elemente -­‐ Strain-­‐Based Criterion Steps to building an FE model of the axon:
Add axoplasm and myelin as solid elements: the
Ranvier node (~1 μm wide) is left unsheathed.
Axoplasm
1011-­‐1016 Elemente -­‐ Axonal Damage Cytoskeleton
Membrane
Ranvier
node
Myelin
8/15/2014
Building a Finite Element Model of Axonal Microscopic Structure
8
Variabilitäten
Variabilität in den Materialien Variabilität in der numerischen Lösung Variabilität in den Randbedingungen
Variabilitäten
Stochastic approach
Eine SimulaUon ist nicht genug um die Biomechanik des Menschen zu verstehen oder zu erklären!
Each Sample/Design
Input parameter
Quelle: Kethu, CAE Grand Challenge, 2012
(Realistic Distributions)
Output parameter
lue
Variabilitäten
Inhalt
Crash Vic6m Simula6on Status der Modelle Möglichkeiten und Grenzen der Verletzungsvorhersage Zukünhige Entwicklungen
Individualisierung
Photo: Howard Schatz
Individualisierung
Method Overview
Parametric
Human Model
Skeleton Geometry Models
BMI=25
Sitting Posture Model
Baseline Models
BMI=30
Mesh
Morphing
External Body Surface Model
BMI=35
BMI=40
Quelle: UMTRI, M. Reed
Individualisierung
Results – Body Scans
Hybrid-III ATD
Quelle: UMTRI, M. Reed
Low-­‐G Anwendungen
Quelle: Brolin, HuMoSym 2014
Pre-­‐Crash Szenarien Heckaufprall of active muscles
in an HBM developed for crash simulations
alidation in the low g loading regime.
Fahrdynamik
Modell Auswahl
Welches Modell kann meine Fragestellung beantworten? Wie ist der Stand der Valida6on? Wer hat das Modell entwickelt? In welchen Programm/Version läuD das Modell?
Modell Auswahl über Verletzungen (AIS-­‐codiert)
Zusammenfassung
Menschmodelle können Verletzungen beim Unfall voraussagen Simula6on mit Menschmodellen sind noch keine industrielle Anwendung Zusammenarbeit in der Entwicklung und Transparenz der ModelleigenschaDen sind notwendig
Vielen Dank.
www.carhs.de