This is a challenging subject to teach

because the domain is so broad and deep

Modern finite element Finite
code (e.g. Abaqus, LSDyna) Element
Methods

Advanced Linear
Programming Algebra
(parallel computing) (equation solvers)

Continuum
Theory of Theory of Theory of
Mechanics
Beams Plates Shells
(Elasticity)

Elementary Mechanics Structural Structural Computer

Statics
Physics of Materials Analysis Dynamics Language
We will draw selectively from the foundational
elements to develop general understanding
Modern finite element Finite Things we will
code (e.g. Abaqus, LSDyna) Element
draw upon
Methods

Advanced Linear
Programming Algebra
(parallel computing) (equation solvers)

Continuum
Theory of Theory of Theory of
Mechanics
Beams Plates Shells
(Elasticity)

Elementary Mechanics Structural Structural Computer

Statics
Physics of Materials Analysis Dynamics Language
3
 How we solve engineering problems

Analytical Numerical

Idealized model Rigorous model

Beam on elastic e.g. Continuum model
foundation of beam and soil

Elastic
or
Inelastic

Exact analytical Numerical (approximate)

solution of governing solution of governing
ODEs and PDEs Solution ODEs and PDEs
 Analytical versus numerical

Analytical
𝑞(𝑥) Governing ODE or PDE Analytical Solution

𝑑#𝑤 𝑥
𝐸𝐼 # =𝑞 𝑥 𝑤 𝑥 = 𝑎𝑥 ! + 𝑏𝑥 " + 𝑐𝑥 + 𝑑
𝑑𝑥

Numerical Discretized structure (FEM) Numerical Solution

System of simultaneous
equations

𝐾 𝑥 = 𝑃

𝑥 = 𝐾 !" 𝑃
Our simulation objective, these days, is to
create a “virtual twin” of the actual system

Actual System

“Virtual Twin”

Desired outcomes for engineers in research and practitioner environments…

- Enhanced understanding of system response

- Quantitative insight on forces, displacements etc.
 Progression of simulation capabilities –
60 years of tremendous advancement
Linear Nonlinear Multi-Physics

Physics-Based Models

DYNA3D

1 1960’s – 1980’s 2 1980’s – 2000’s 3 2000’s – 2020’s

Nonlinear Analysis Multi-Physics
Low Fidelity Linear
- Material Nonlinearities
Engineering Mechanics Applied Math / Numerical Methods
Elastic Analysis David Floriana Anders Bjorn Wei
McCallen Petrone Petersson Sjogreen Liu
- Geometric Nonlinearities
(contact)
Clough
1956 Computer Science Seismology / Geophysics
DYNA3D
Houjun Ramesh Arben Arthur Rie
LSDYNA Tang Pankajakshan Pitarka Rodgers Nakata

OpenSees
Mckenna Hallquist

SAP SASSI
Wilson Lysmer Hero Programmers Code Teams
 The three foundational elements
of numerical simulations
Underlying physics
Discretization of Numerical
and governing
the system solution
mathematical equations
(FEM / FD) schema
(ODE’s and PDE’s)
Element level
Physical
component

Engineering theory
Assembly

System level

Governing equations
 Foundational elements – theory and
governing equations (ODE’s and PDE’s)
Solid (Continuum) Mechanics
e.g. theory of elasticity
Soil Concrete Metal

Scientists Engineers

Phenomenological
models
 Foundational elements – theory and
governing equations (ODE’s and PDE’s)
Engineering Theories

Beams

Plates

Shells
 Discretization technologies (FEM) are pretty
well established from years of research
Solids Beams

Plates Shells
 Numerical simulation schemas are
tailored to the problem at hand
Crash

Explicit time integration Event duration 0.140 seconds

Implicit time integration
Earthquake

Transient Interstory
Drift (TID(t))

Residual Interstory
Drift (RID)

Landers near-field motions

Peak Interstory
Drift (PID)

Event duration 40-100+ seconds

 An expert modeler must be knowledgeable
and experienced in modeling systems
Underlying physics
Discretization of Numerical
and governing
the system solution
mathematical equations
(FEM / FD) schema
(ODE’s and PDE’s)

1) Develop a modeling 2. Create a 3) Convince yourself the

strategy representative model model is working
What physical model (reality check)
Appropriate discretization
represents the problem, and ability to refine the
because a model topologically Run simple “diagnostic”
model simulations, e.g.
”looks” like the real system
doesn’t make it right - Simple push-over
- Compare to simple models
- Eigen solution etc.
- Exploit visualization

San Francisco Bay Bridge deck

An expert modeler must have systematic best
practices to avoid self-induced faults

Minimize the potential for “pilot error”

Numerical
solution
schema

3) Convince yourself the

model is working
(reality check)
Run simple “diagnostic”
simulations, e.g.
- Wrong theory
- Simple push-over
- Units wrong / mixed
- Compare to simple models
- Eigen solution etc.
- Error in mesh
- Exploit visualization - Error in element connections
- Boundary condition errors
on and on….
 You must know the limitations in your
theory and computational models!
Engineer’s Key Questions
objective…
Physical system • What is the appropriate physical &
“Virtual twin” mathematical model idealization?
- beam theory for frame elements
- plate theory for floor slabs
- shell theory for shear walls

• What is the appropriate

discretization / numerical strategy?
- select beam, plate, shell finite elements
- determine necessary discretization
- static or dynamic solution schema

• What is the range of applicability of

the model I have created?
- onset of nonlinear behavior!
 What basic assumptions are implicit
in a traditional linear elastic approach?

Material behavior is linear elastic

s F
Loading

Unloading

e D
System response is not dependent
on the system displacements

H d<H
d
 What are some system features
that can invalidate a linear model?

Nonlinear material behavior

s s

Loading

Unloading

e e
Geometry dependent behavior F
(finite deformation)

D
 Nonlinear simulations for risk-informed
performance-based design (auto crash)

More frequent small crunches Rare large crashes

Sacrifice the car

Sacrifice the bumper (relatively low $)
Protect the occupants
Protect the body/frame (very high $)
Optimal design demands
LSDYNA
realistic simulation models!
 Nonlinear simulations for risk-informed
performance-based design (earthquakes)
More frequent
smaller earthquakes
Small
No Damage
Damage
Less frequent
big earthquakes

Optimal design demands

realistic simulation models!

Rare or extreme earthquakes

Frequent small earthquakes protect life safety and
little or no damage potentially functionality
We now have nonlinear element technologies
for steel and concrete structures

W
Steel σ

36,000 psi

120 inches length

W 14 x 159 cross section

36,000 psi Ey = E/5

20

15

10

Force [kN]
Concrete 0

-5

-10

-15 ESSI
OpenSees
-20
-200 -100 0 100 200
Displacement [mm]
 An effective simulation ecosystem
has three equally strong elements

Model / mesh Numerical Post

generation simulation processing
 Model generation is often the most
time-intensive element of simulation
For complex systems, animated analysis is
the only way to gain full understanding

23
 Is the simulation model representative of
the system you are modeling? - (V&V)

• Verification—Verification is the process of confirming that

a computer code correctly implements the algorithms that
were intended. The process of confirming that the equations
are numerically solved accurately.

• Validation—Validation is the process of confirming that

the predictions of a code adequately represent measured
physical phenomena. The process of confirming that the
equations are (physically) accurate.

To the first order this is the responsibility of the code developer,

however, the user may have do V&V for selected unique problems
 Comparison to experimental testing
is often essential to strong validation

21.5 ft.
15 ft.

In the UNR earthquake laboratory

we are building a major
new experimental capability
for validating an advanced
nonlinear code for
Soil-Structure-Interaction
 Application example - ground motions and
building response in the near-field region
Basin B

Hypocenter

A
Strike-slip Sedimentary
Fault Basin (600m)
Slip (cm)
Graves and Pitarka rupture model0 / 119 / 471
m7.00−62.5x16.0_s200−G9_3_v5.5.0_HB
0
600

5
W (km)

400

10
200

15
0

Section A - B 0
Rise Time (s) 0/1/3
3.0
 Visualization engine - 40 story steel frame
response in the near-field (strong coupling)
9 billion zone SW4
geophysics wave
propagation model

Fault normal displacement, velocity and acceleration

Fault
normal

Fault
parallel
ASCE 43-05 Limit States

0 0.5% 1.0% 2.5% 3.5%

 Regardless of your career path, a grasp of
simulation will be a tremendous asset!
Research & Development Engineering Practice

Architect’s
Shear wall concept
experiment
(complex
shear-wall
frame system)

Shear wall d
P

simulation
Engineering
reality

Deep understanding!
 I hope you all will develop more
appreciation for the “magic” of simulation

Physical
system Hardware Voila!
Software
- A system of commands

Storing data in memory

Set of mathematical
operations and
algorithms Binary arithmetic