Islamic University of Technology
Biomedical Engineering
ME 4873
Lecture 06
Modelling and simulation of biological systems
M Nasim
Introduction
v Human body models, primarily
developed for crash analyses, can
be subdivided into
i. models of crash dummies and
ii. models of real human bodies.
v They account for inertial effects by
somehow deriving equations of
motions for all movable parts and
solve these equations by a (a) Hybrid III model [adapted from Eliasson and Wass,
numerical method. 2015], (b) GHBMC model [adapted from GHBMC.com,
2018], and (c) THUMS model [adapted from
Dynamore.com, 2018]
M Nasim | 1
Mathematical Models
v In the crash safety field, mathematical models can be applied in practically all
area's of research and development including:
o crash response of vehicles, safety devices and roadside facilities.
o reconstruction of actual accidents
o human impact biomechanics studies
v The mathematical formulations used for these models can be subdivided into
i. lumped mass models,
ii. multi-body models, and
iii. finite element models.
M Nasim | 2
Lumped-mass model
v In a lumped mass model, a system is represented
by one or more rigid elements often connected by
mass-less elements like springs and dampers.
v The model shown in the figure simulates the thorax
response in case of a loading by an impactor, which
consists of 3 rigid bodies connected by springs and
dampers.
v One mass represents the impactor mass and other
masses the sternal and vertebral effective mass,
respectively.
v Springs and dampers represent the skin and flesh
between impactor and sternum and the connection
between sternum and thoracic spine. Thorax model [adapted from Lobdell
et al., 1973]
M Nasim | 3
Multi-body (MB) Human Body Models
v The most important difference between a
lumped mass model and a multi- body
model is that bodies in a multi-body
formulation can be connected by various
joint types, due to which the number of
degrees of freedom between the
elements can be constrained.
v Another difference with lumped mass
models is that in a multi-body formulation,
instead of rigid bodies, also flexible
bodies can be specified.
(a) 5th percentile MADYMO female model [adapted
from Xu et al. 2016] and (b) 50th percentile MADYMO
male model [adapted from Anderson et al., 2007]
M Nasim | 4
Finite Element (FE) Human Body Models
v The finite element method is a
numerical technique to solve equilibrium
equations for a domain with arbitrary
shape and constitutive properties.
v The system to be modelled is divided in
a number of finite volumes, surfaces or
lines.
v These elements are interconnected at a
discrete number of points: the nodes to
which degrees of freedom are
associated.
v The state of stress follows from the
THUMS model [Dynamore.com, 2018]
deformations and the constitutive
properties of the material modelled.
M Nasim | 5
Example of FE models: Head-neck Complex
[Courtesy: Toyota Motor Corporation] M Nasim | 6
FE Models for Various Applications
[Courtesy: Toyota Motor Corporation] M Nasim | 7
Model Generation
// What is Wrong?
"All models are wrong, but some are useful”
- George E.P. Box, 1976
// Nobody believes the model, except the modeler.
// Everybody believes the experiment, except the experimentalist.
M Nasim | 8
Development Phases
Pre-processing Solver Post-processing
Material
ti o n
properties i da
Val
Geometry - // Implicit
Model Convergence
Mesh // Explicit
Inju
ry P
Boundary re d
ictio
conditions n
M Nasim | 9
Geometry
The geometry of computational
model can be achieved by:
// Model refinement
// Anatomical structure
// MRI
// CT scan
// CAD
M Nasim | 10
Mesh Generation
// Discretization
1D, 2D shell, 3D solid
Discrete Triangular Rectangular Hexahedral Tetra
M Nasim | 11
Mesh Generation
// Discretization
1D, 2D shell, 3D solid
// Element size
M Nasim | 12
Mesh Generation
// Discretization
1D, 2D shell, 3D solid
// Element size
// Element quality
// Element formulation
M Nasim | 13
Example 1: FE Neck Model development
3D CAD design of cervical vertebra
[Nasim et al., 2022] M Nasim | 14
Example 1: FE Neck Model development
Mesh generation
[Nasim et al., 2022] M Nasim | 15
Example 1: FE Neck Model development
Intervertebral discs generation and adjustment
[Nasim et al., 2022] M Nasim | 16
Example 1: FE Neck Model development
Spinal cord and cartilage adjustment
[Nasim et al., 2022] M Nasim | 17
Example 1: FE Neck Model development
Coupling of the neck model with a head model
[Nasim et al., 2022] M Nasim | 18
Example 1: FE Neck Model development
Generation of 1D elements for ligaments Generation of 1D elements for muscles
[Nasim et al., 2022] M Nasim | 19
Example 1: A Simple 3D FE Neck Model
Complete head-neck complex with addition of the skin and fleshy tissue of the neck
[Nasim et al., 2022] M Nasim | 20
Example 2: FE Head Model development
// Series of MRI/CT scan images
// Selection of the tissues at each image
// Construction of the 3D model as a
summation of all the images
M Nasim | 21
Example 2: Detailed Head Model development
Head model: (a) 3D view, (b) cross-sectional view in
sagittal plane [adapted from Pavan et al., 2022]
M Nasim | 22
Material properties
// Simulations with human body finite element models require constitutive
descriptions for the various materials that constitute the human body (such as
bone, cartilage, ligament and tendon, muscular tissue, and various organs such
as brain, heart, lung, etc.).
// Accurate representations of the mechanical behaviour of various components
are essential for biofidelic response of the models.
M Nasim | 23
Material properties
// Most biological materials exhibit non-linear behavior
// Also sensitive to strain rate
// Assigned by a suitable material model
// Defined by constitutive laws of stress and strain,
M Nasim | 24
Convergence
// Convergence is a crucial aspect of FEA, which
refers to the state of the computed solution -
where it becomes stable and does not change
significantly with further refinement of the
numerical model parameters such as mesh
density or iteration steps.
// Choosing the correct mesh size
// Mesh size depends on
> Material properties and failure
> Boundary conditions
> Loading
> Contacts
> Assessment metric
M Nasim | 25
Validation
// Validation is the process of assessing the reliability of a simulation model in
comparison to one or more reference tests with human subjects.
// Very important in this process is that the reference tests, often also referred to
in literature as biofidelity tests. These tests are not the same as used to provide
model input data.
// If results from different tests are available, usually so-called biofidelity corridors
are defined for instance consisting of envelopes of resulting time histories.
M Nasim | 26
Validation: Boundary Conditions
// Boundary conditions define how a
system (for example, a structure or a
fluid) interacts with the environment.
// Fixations, loads, pressures, flow rates,
or velocities are all examples of
boundary conditions.
// Before solving any computation model,
the boundary conditions in the
simulation must be similar to that in the
experimental tests used for the (a) Experimental test setup [adapted from Nightingale
validation. et al. 2016] and (b) FE model for simulation [adapted
from Nasim et al., 2022]
M Nasim | 27
Validation: Qualitative Analysis
// Qualitative
> Kinematics
[Nasim et al., 2022] M Nasim | 28
Validation: Qualitative Analysis
// Qualitative
> Kinematics
> Response
[Nasim et al., 2022] M Nasim | 29
Validation: Quantitative Analysis
// Qualitative
> Kinematics
> Response
// Quantitative
> Single/multiple parameter
> Corridor (ISO)
> Cross-correlation (CORA)
[Nasim et al., 2022] M Nasim | 30
Validation: Quantitative Analysis
// Qualitative
> Kinematics
> Response
// Quantitative
> Single/multiple parameter
> Corridor (ISO)
> Cross-correlation (CORA)
Based on CORA score: the response
of the model exhibits ‘Good Biofidelity’
[Nasim et al., 2022] M Nasim | 31
Models’ Offers
// Interpretation of experimental results
// Investigative response to impact (sensitivity studies)
// Considering new designs for protection and safety
// An approximate representation of a physical phenomenon, bounded by their
assumptions and have a finite life.
M Nasim | 32
Application: Injury Prediction
M Nasim | 33
Application: Injury Prediction Example
[Courtesy: Lotta Jakobsson; Volvo Cars] M Nasim | 34
Application: Injury Prediction Example
[Courtesy: Toyota Motor Corporation] M Nasim | 35
Application: Accident Reconstruction
Car to Pedestrian Car to Car Car to Bicycle Car to Motorcycle
[Courtesy: Toyota Motor Corporation] M Nasim | 36
Application: Automated Emergency Braking
[Courtesy: Toyota Motor Corporation] M Nasim | 37
Application: Autonomous Vehicle Safety
[Courtesy: Toyota Motor Corporation] M Nasim | 38
Application: Segmented Parts Analyses
[Courtesy: Toyota Motor Corporation] M Nasim | 39
Application: Helmet Testing
[adapted from Siegkas, 2019]
M Nasim | 40
