Technical Manual

Extreme Loading for Structures
Technical Manual
Extreme Loading is made possible by

Copyright January 2006
Extreme Loading for Structures Technical Manual
COPYRIGHT
The software program Extreme Loading for Structures and all
associated documentation are proprietary and copyrighted
products. Worldwide rights of ownership rest with Applied
Science International. Unlicensed use of the program or
reproduction of the documentation in any form, without prior
written authorization form Applied Science International is
prohibited.
Further information and copies of this documentation may be
obtained from:
Applied Science International

The Steel Network, Inc. 3221 Wellington Ct., Raleigh, NC
USA 27615
National Sales and Technical Support
Phone: (919) 645-4090
Fax:
(919) 645-4085
E-Mail
technicalsupport@extremeloading.com
customersupport@extremeloading.com
www.extremeloading.com
2006 Applied Science International, LLC
DISCLAIMER
Extreme Loading for Structures, Version 1, License Agreement upon Installation:
Read this License Agreement carefully. By installing, copying or otherwise using this software
product, you agree to be bound by the terms below. If you do not agree to these terms, ASI
("ASI") is unwilling to license the Software to you. If you do not agree to the terms, promptly
return to ASI the Software and all accompanying written documentation. "Software" as used
herein includes the computer software, the associated media, any printed materials and any online or electronic documentation.
License: The Software is licensed only, not sold. ASI grants you the right to use the Software for
internal business purposes only. ASI may, from time to time, provide you with updates or
revisions to the Software, but is not obligated to do so under this License Agreement. The terms
of this License Agreement shall govern any updates or revisions to software provided by ASI
which is not subject to a separate license agreement.
Restrictions: You acknowledge and agree that the Software is proprietary and is protected by
copyright and other laws. You may not remove any copyright or other proprietary notices from
the Software and shall include such notices on any copies made by you. You may not use the
Software on behalf of or in service to other parties other than for your own internal business
purposes. You acknowledge that the Software contains trade secrets. You may not under any
circumstances alter, modify, adapt, translate or create a derivative work based on the Software,
or disassemble, decompile or otherwise reverse engineer the Software.
Ownership: ASI owns all right, title and interest in and to the Software, including all intellectual
property rights and reserves all rights not expressly granted hereunder. This License Agreement
does not convey to you an interest in or to the Software, but only a limited right of use revocable
in accordance with the terms of this License Agreement.
Maintenance Service: If you have purchased maintenance and support services for the Software,
ASI will provide such services for an initial term of one year. You may elect to renew coverage
for each successive year by paying the then applicable fee for such services. You will be invoiced
for the applicable fee for each renewal maintenance period at least thirty days prior to the
commencement of such maintenance period. If the applicable fee is not paid prior to the
commencement of such period, you will be considered to have terminated the maintenance and
support coverage for the Software. ASI will have no obligation to provide services on account of
any problems caused by any unauthorized modifications to the Software. ASI reserves the right
to limit maintenance services to the then current version of the Software. All services provided
under this section and any and all upgrades, revisions or enhancements to the Software will be
governed by and subject to this License Agreement.
Termination: Without prejudice to any other rights, ASI may terminate the license if you violate
any of the terms of this Agreement. In such event, you must immediately cease all use and
destroy the original and all copies of the Software.
Limited Warranty: ASI does not warrant that the Software will meet your requirements or that
operation of the Software will be uninterrupted or error-free. As your sole remedy upon a breach
of this warranty, ASI will, at its option, replace any defective media. The Software is provided to
you as a preliminary design development tool. The results are not, and should not be treated as,
a replacement for the judgment of design professionals of record. All information and results
produced by the Software should be reviewed by the licensed design professionals of record on
the specific project and for the intended use.

THE FOREGOING WARRANTY IS IN LIEU OF ALL OTHER WARRANTIES, WHETHER
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
PURPOSE. THE FOREGOING STATES ASI's ENTIRE LIABILITY AND YOUR EXCLUSIVE
REMEDY. IN NO EVENT SHALL ASI BE LIABLE FOR SPECIAL, INCIDENTAL,
CONSEQUENTIAL, INDIRECT OR SIMILAR DAMAGES
ARISING OUT OF THE USE OF OR INABILITY TO USE THIS PRODUCT INCLUDING
DAMAGES FOR LOSS OF PROFITS, BUSINESS INTERRUPTION, LOSS OF DATA OR THE
LIKE), EVEN IF ASI HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. IN NO
EVENT WILL ASI'S LIABILITY EXCEED THE AMOUNT YOU PAID FOR THIS PRODUCT.
SOME STATES DO NOT ALLOW THE EXCLUSION OR LIMITATION OF IMPLIED
WARRANTIES OR LIMITATION OF LIABILITY FOR INCIDENTAL OR CONSEQUENTIAL
DAMAGES, SO THE ABOVE LIMITATIONS MAY NOT APPLY TO YOU.
Other Software and Trademarks: The software of other parties used in connection with the
Software remains the property and responsibility of the other parties and not ASI. MS Software
Products is a registered trademark of Microsoft Corporation. AutoCAD is a registered trademark
of Autodesk Inc. No right, license or interest in any trademark of ASI or others is granted
hereunder, and you agree that no such right, license, or interest shall be asserted by you with
respect to any such trademark.
Governing Law: This Agreement is governed by and shall be construed in accordance with the
laws of North Carolina. Should any provision of this License Agreement be declared void or
unenforceable by any court of competent jurisdiction, such declaration shall have no effect on the
remaining provisions hereof.
U.S. Government Restricted Rights: The Software and related documentation are provided with
RESTRICTED RIGHTS. Use, duplication, or disclosure by the Government is subject to
restrictions as set forth in subparagraph (c)(1)(ii) of the Rights in Technical Data and Computer
Software clause at DFARS 252.227 - 7013 or subparagraphs (c)(1) and (2) of the Commercial
Computer Software - Restricted Rights at 48 CFR 52.227-19, as applicable. Manufacturer for
such purpose is ASI.
If you have questions concerning this Agreement or wish to make a warranty claim, please write
to:
Address:
3221 Wellington Court,
Raleigh, NC 27615
USA
Phone: (919) 645-4090
Fax: (919) 645-4085
Email: asi@appliedscienceint.com
Table of Contents
1
INTRODUCTION............................................................................................................................... 1
APPLIED ELEMENT OVERVIEW................................................................................................. 5

2.1
SHAPE .......................................................................................................................................... 5
2.2
CONNECTIVITY ............................................................................................................................. 6
2.2.1
Element connectivity at FEM.................................................................................................. 6
2.2.2
Element connectivity at AEM.................................................................................................. 9
2.3
DEGREES OF FREEDOM (DOF) ................................................................................................... 11
2.4
SUPPORTS ................................................................................................................................... 12
2.4.1
Rigid Supports ...................................................................................................................... 12
2.4.2
Planes Of Symmetry And Skew Symmetry ............................................................................ 14
2.4.3
Intersection Between Planes Of Symmetry ........................................................................... 16
2.5
UNITS ......................................................................................................................................... 16
2.6
MATERIAL PROPERTIES .............................................................................................................. 16
2.6.1
Youngs Modulus .................................................................................................................. 16
2.6.2
Shear Modulus...................................................................................................................... 17
2.6.3
Tensile Strength (Or Tensile Yield for Steel) ........................................................................ 18
2.6.4
Compressive Strength (Or Compressive Yield for Steel) ...................................................... 19
2.6.5
Ultimate strength/ yield stress .............................................................................................. 19
2.6.6
Separation Strain.................................................................................................................. 20
2.6.7
Friction Coefficient............................................................................................................... 21
2.6.8
Specific Mass ........................................................................................................................ 22
2.6.9
Damping Ratio r............................................................................................................... 22
2.6.10
Post-Yield Stiffness Ratio ................................................................................................. 23
2.6.11
Contact Parameters ......................................................................................................... 24
2.7
CONNECTIVITY SPRINGS ............................................................................................................ 26
2.7.1
Matrix Springs ...................................................................................................................... 26
2.7.2
Reinforcement Springs.......................................................................................................... 27
2.8
ANALYSIS .................................................................................................................................. 28
2.8.1
Direct Solver......................................................................................................................... 29
2.8.2
Iterative Solver ..................................................................................................................... 31
MATERIAL MODELS AND FAILURE CRITERIA ................................................................... 32

3.1
3.2
3.2.1
3.2.2
3.2.3
3.3
3.4
LOADING ......................................................................................................................................... 39
4.1
ELASTIC MODEL ........................................................................................................................ 32

REINFORCED CONCRETE MODELS .............................................................................................. 32
Concrete models ................................................................................................................... 32
Reinforcing bars Model ........................................................................................................ 32
Cracking Criterion Of Concrete ........................................................................................... 33
COMPOSITE STEEL-CONCRETE MODELS .................................................................................... 34
MASONRY WALL MODELS ......................................................................................................... 36
WHAT ARE LOADING STAGES?.................................................................................................... 39
LOADING TYPES............................................................................................................................ 41
5.1
STATIC LOADING ....................................................................................................................... 41
5.1.1
Own Weight .......................................................................................................................... 41
5.1.2
Lumped Weight (mass) ......................................................................................................... 41
5.1.3
Load Control ........................................................................................................................ 42
5.1.4
Displacement Control........................................................................................................... 42
5.2
DYNAMIC LOADING ................................................................................................................... 43
5.2.1
Calculation Time Step........................................................................................................... 43

5.2.2
5.2.3
5.2.4
5.2.5
5.2.6
5.2.7
5.2.8
6
BOUNDARY AND INITIAL CONDITIONS................................................................................. 51

6.1
6.2
6.3
6.4
INTRODUCTION........................................................................................................................... 55
CONTACT TYPES ........................................................................................................................ 55
Corner-Face type.................................................................................................................. 55
Edge-Edge Type.................................................................................................................... 56
Corner-Ground Type ............................................................................................................ 56
CONTACT STIFFNESS .................................................................................................................. 56
ENERGY DISSIPATION DURING CONTACT .................................................................................. 57
TIME STEP .................................................................................................................................. 59
COMMON MODELING TRICKS ................................................................................................. 60

8.1
8.2
8.2.1
8.2.2
8.3
8.3.1
8.3.2
8.4
8.5
8.6
8.7
ANALYSIS DOMAIN BOUNDARIES .............................................................................................. 51

ASSIGNING BOUNDARIES IN ELS ............................................................................................... 51
ASSIGNING BOUNDARY CONDITIONS (RESTRAINS OR SUPPORTS).............................................. 52
INITIAL CONDITIONS (VELOCITY OR ACCELERATION) ................................................................ 52
ELEMENTS CONTACT ................................................................................................................. 55

7.1
7.2
7.2.1
7.2.2
7.2.3
7.3
7.4
7.5
Load Control ........................................................................................................................ 44

Displacement Control........................................................................................................... 44
Earthquake ........................................................................................................................... 46
Inertia Loads from Lumped Mass......................................................................................... 46
Blast Scenario....................................................................................................................... 47
Element Removal .................................................................................................................. 48
Pressure Control................................................................................................................... 49
CONTACT PROBLEM GUIDANCE ................................................................................................. 60

COMMON STABILITY PROBLEMS ................................................................................................. 60
Static Loading....................................................................................................................... 60
Dynamic Loading ................................................................................................................. 61
GUIDANCE FOR MODELING REINFORCING BARS ......................................................................... 61
Reinforcing Bars Extension Length ...................................................................................... 61
Modeling Of Non Symmetric Bars In Columns..................................................................... 62
PROPERLY INCLUDING OWN WEIGHT ........................................................................................ 63
POST-PEAK BEHAVIOR ACHIEVEMENT ...................................................................................... 63
CONTACT SPRINGS UNLOADING STIFFNESS FACTOR ................................................................. 64
CYCLIC POINT LOADING ............................................................................................................. 64
VERIFICATION SAMPLES........................................................................................................... 66
9.1
STATIC LOADING ....................................................................................................................... 66
9.1.1
Elastic stage.......................................................................................................................... 66
9.1.2
Non-linear Stage................................................................................................................... 66
9.2
DYNAMIC LOADING .................................................................................................................... 68
9.2.1
Elastic Stage ......................................................................................................................... 68
9.2.2
Non-linear Stage................................................................................................................... 68
9.2.3 .................................................................................................................................................... 68
9.2.4
Demolition and Blast ............................................................................................................ 69
10
CASE STUDIES................................................................................................................................ 70
11
REFERENCES.................................................................................................................................. 71
1 Introduction
Structural analysis and design has been practiced using the finite-element method for
over five decades. Design code, in most cases, allow the use of linear elastic analysis to
calculate internal forces in each element. Design procedures account for internal force
values, with safety factors, to return applicable concrete dimensions, reinforcement areas
and estimated deflection and serviceability assurance that will ensure the safety of the
structure for its intended use.
Previous approximate analysis and design techniques are adequate with structures
subjected to traditional vertical gravitational loads and earthquake equivalent static loads.
In brief, approximate analysis methods may work fine when the purpose is to design
structural elements to withstand the applied loads. However, a structure may be subjected
during its lifetime to extreme loading conditions that exceed its design loads. Amongst
these loading conditions are major earthquakes, blast-effects, unexpected impact forces,
and fire. Unfortunately, many structures are not being designed to resist such extreme
loads due to economic reasons. Life safety considerations necessitate that in the event of
an extreme loading condition on a building, people can be evacuated safely before the
building collapses. This requires a forecast about whether a building would eventually
collapse in such an event or not, and such a forecast is applicable for future structures as
well as existing structures. By reviewing the casualties caused by previous major
earthquakes around the world, it was found that more than 90% of the death toll was due
to structural collapse of buildings and bridges. A forecast that the towers of the World
Trade Center would collapse from the extreme impact load and fire resulting from the
plane crashes on September 11th, 2001 may have saved thousands of human lives. The
question that arises here is that, can we use approximate analysis methods to simulate
such critical structural behavior upon which the safety of thousands of people will be
decided?
Computer simulation is an important key in determining the performance of structures in
extreme loading conditions. However, it is not possible to predict the beavior of collapse
for structures using the finite-element method Reviewing current research, it can be
noticed that methods used for structural analysis are mainly based on rules of continuum
mechanics, like the finite-element method (FEM), which cannot be applied explicitly to
discrete elements. Therefore continuum mechanics-based methods cannot be extended to
simulate the collapse analysis. On the other hand, analysis methods based on rules of
discrete material cannot be used to predict behavior of continuum elements. As a matter
of fact, structures during a collapse situation pass through the two stages; a continuum
stage first, and then a discrete stage. The analysis and simulation needs to follow both
behavior stages in order to help in answering the following questions:
- Will the structure collapse in an extreme loading event?
- Will the collapse be partial or total collapse?
- What is the mode of collapse of the structure?
- In cases of partial collapse, will it be possible to repair the structure?
- How long would it take for the structure to get to completely collapse?
- How does falling debris affect adjacent structures?

- What is the footprint affected by the collapse of a structure?
These questions are a small sample of questions that cannot be answered without having
an accurate prediction of the structural performance when subjected to an extreme
loading.
A new method, that is capable of predicting to a higher degree of accuracy the continuum
and discrete behavior of structures, has been developed. Through nine years of
continuous development, the Applied Element Method (AEM) was proved to be the
method that can track the structural collapse behavior passing through all stages of the
application of loads, elastic stage, crack initiation and propagation in tension-weak
materials, reinforcement yielding, element separation, element collision (contact), and
collision with the ground and with adjacent structures. The possible analysis domain of
AEM comparable to FEM is shown in Fig. 1.1. International publications verify that the
AEM can cover with a reasonable accuracy the fields of application presented in Table
1.1 where the yellow bar represents continuous research and development.
Accurate
Accurate
Nonlinear
Geometrical &
Material Changes
Element
Separation
Reliable Results
Reliable Results
Fig. 1.1 Analysis domain of AEM
Collision
Progressive
Collapse
AEM
Elastic
Large Displacement
FEM
Small Displacement
Table 1.1 Application fields of AEM
Static
Geometry
Small
displacements
(linear)
Large
displacements
(nonlinear)
Material
Monotonic
Elastic
Reference No. 2
Nonlinear
Dynamics
Cyclic
Reference No.
1
Reference No. 3
Reference No. 4, 6, 7
Elastic
Reference 5
Nonlinear
Covered in dynamics
Covered in dynamics
Collapse Process
The method accuracy is compared with more than 50 experimental and theoretical
results. More details about the verification samples are presented at
www.extremeloading.com.
Literature surveys indicate that no other simulation technology has comparable overall
performance to AEM.
Applied Element Method (AEM) is the base method to analysis of structures under
extreme loadings. This table shows a brief comparison between the Applied Element
Method (AEM) and the Finite Element Method (FEM) and you can easily judge why
AEM was used for the structural analysis:
Table 1.2 Comparisons between AEM and FEM
CPU time
AEM
FEM
Short
Short
Degrees of freedom 6 per element

Cracking model
16 per element (8-node element)
Smeared cracks and physical

cracks
Smeared cracks and physical

cracks (interface cracks, whose
locations should be predefined)
No need for joint elements
Joint elements (interface

elements) are needed at wide
cracks' locations for simulating
physical cracks
No need for joint elements
Joint elements (interface

elements) location should be
predefined before analysis
Cracks propagates easily at

element boundaries in any
direction
Cracks can not extend from

element to element which makes
the structure inseparable at
arbitrary locations
Crack propagation can be

followed
Preprocessing time Short
Crack propagation cannot be

followed accurately in smeared
crack zones
Long
Reinforcement
details
All reinforcement details, for

each reinforcement bar is
accounted for. For example, bar
area and concrete cover can be
taken into account.
Very difficult to accurately

account for reinforcement details
in accurate way as properties
should be added at integration
point locations.
Before collapse
High Accuracy
High Accuracy
During and after

collapse
High Accuracy
Can not follow collapse behavior
2 Applied Element Overview

With AEM, the structure is modeled as an assembly of small elements, which are made
by dividing the structure virtually, as shown in Fig. 2.1-a. The two elements shown in
Fig. 2.1 are assumed to be connected by one normal and two shear springs located at
contact points, which are distributed around the elements edges. Each group of springs
completely represent stresses and deformations of a certain volume as shown in Fig. 2.1b.
Volume represented by a
normal spring and 2 shear
springs
Reinforcement bar
Structure boundary
a. Element generation for

AEM
b. Spring distribution and area of influence of each

pair of springs
Fig. 2.1 Modeling of structure to AEM
The following sections provide an overview about the properties of an Applied Element.
2.1 Shape
In most commercial finite-element codes, Frame elements are used to model columns
and girders. These frame elements are mainly lines connecting two joints. Each line has
properties of area and inertia. It should be emphasized that using frame elements in FEM
structural analysis is suitable and sufficient for conventional structural design. However,
it will not work with structural collapse behavior. Such codes can not have contacts
between failed elements if it is modeled using frames. Moreover, the collapsed frames
themselves do not keep their shape after collapse as they break down into small pieces.
Therefore, it would not be convenient to use frame elements in progressive collapse
analysis. In most of the FEM codes, plates are modeled as shell elements where the
element thickness is an element property. This is also an obstacle for the progressive
collapse of slabs. In few advanced nonlinear FEM codes, the frame elements are modeled
using 3D solid elements, which represent an accurate way of calculating stresses in
concrete and detecting the realistic failure mechanisms. However, it still can not simulate
a real collapse as the solid elements are compatible in deformation along with their
intersection nodes and can not be separated.
In AEM, each element has 3-D physical coordinates and shape. Hence, elements are not
just lines or shells but a group of 3D elements which can be separated and or collided
together. In 3-D AEM analysis, cuboids are used to model the structure to be analyzed.
One or more faces can be collapsed into a point or line and hence, 4-point, 5-point, 6
5
points pyramids or prisms can be formed. Each element center of gravity is calculated at
which degrees of freedom (unknown displacements) are calculated.
When designing the mesh to be analyzed, it is recommended to:
1- Cube-shape elements are optimal elements to be analyzed. However, use of cube
shape in problems where the thickness is very small, as with slabs, will lead to a
large number of elements. Hence, paralelloid-shape elements can be used in this
case with also good accuracy. In brief, the optimal element to use for analysis is
one with perpendicular edges.
2- It is not recommended to have an abrupt change in element size, that is, to have
elements of large size adjacent to elements of small size. A gradual change in
element size is recommended.
3- Use of elements of small sizes is recommended at locations of stress
concentrations. While elements of large sizes can be used at locations of low
stress gradients.
4- For connectivity reasons, element faces should lie exactly at the same plane to
generate connectivity springs between faces. Hence, the user should use snap
options in the graphical user interface to generate elements.
2.2 Connectivity
2.2.1 Element connectivity at FEM
In FEM analysis, elements are usually connected at nodes. Referring to Fig. 2.2, node
number 13 is shared among elements 6, 7, 10 and 11. Hence, all elements should have the
same displacement at this node.
Connectivity methods using FEM works fine when elements are not expected to separate.
However, having element separation, which is the case in all collapse simulation, means
that element displacements should be independent which can not work using the same
joint ID for all elements. Previous researches attempted using multiple node IDs at
expected separation locations. Unfortunately this technique cause stress singularity Very
high stresses at location of nodal separation and hence, this technique did not work.
16
11
10
2
2
14
15
8
9
3
20
12
7
8
7
1
11
13
12
5
19
18
17
9
10
4
Fig. 2.2 Nodal compatibility in FEM
To overcome the above problem of cracking, two techniques are always used in
conjunction with FEM as explained in the following sections.
2.2.1.1 Smeared Crack Approach
In smeared crack approach, the cracks are assumed implicitly inside the element. So,
there is no physical crack but the effect of cracks are considered in the element stiffness
and the material modeling of concrete and reinforcement. Looking to Fig. 2.3, the actual
crack is shown 9 in Fig. 2.3-a. Figure 2.3-b shows how this crack is dealt with in FEM.
This actual crack is represented in the stiffness matrix formulation and the element
nonlinear models of the hatched elements based on assumed crack angle. Here are the
problems from these assumptions:
1- There is no physical crack in the model, the elements are not separated at the
crack location
2- There is no localization of the crack. This means that the crack inside the element
does not have a specific geometry. However, it should be mentioned that, it is still
possible to simulate the crack localization using FEM through reducing the
element size and considering material models that are based on the concrete
fracture energy and the element size, i.e. size-dependant models.
3- No continuity between cracks. Although the cracks are assumed to be continuous
between elements, the FE mesh does not weigh that into the analysis. Referring to
Fig. 2.3-b, the cracks inside each element have its own direction and location and
have no relation with cracks of adjacent elements.
4- Even at the same element, crack orientation at each integration point may be
different
In brief, you cannot see a physical crack to deal with using smeared crack approach.
Hence, you can not separate elements
Actual crack
a- Physical crack
Simulated crack
inside FE elements
b- Simulated cracks in FEM

Fig. 2.3 Smeared crack approach in FEM
2.2.1.2 Discrete Crack Approach

In the discrete crack approach, the crack location should be predefined prior to the
analysis at locations of expected cracks (see Fig. 2.4). If this crack is not included before
analysis, results will not be correct. If predefined crack location is wrong, the element
will be forced to break at that location. The main disadvantages of this technique are:
1- It is totally dependent on the experience of the design engineer. An inexperienced
engineer will produce significantly incorrect results.
2- Setting crack location before analysis may be acceptable in some cases. But in
cases of progressive collapse case where the collapse behavior is unknown before
simulations this is neither acceptable nor practical.
3- Setting predefined crack location before the analysis will force the structure to fail
in a specific way depending on the orientation of discrete cracks
Predefined Discrete Cracks

Fig. 2.4 Discrete crack approach in FEM
2.2.1.3 Moving Mesh Techniques

This technique can be used only for cases with a few cracks (one or two cracks, as shown
in Fig. 2.5) within a brittle material. So, this technique can not be used for real collapse
behavior of structures.
crack
crack
crack
(a)
(b)
crack
(c)
crack
(d)
(e)
Fig. 2.5 moving mesh technique
2.2.2 Element connectivity at AEM

As previously discussed, FEM elements are connected at nodes. Having different node
IDs means that elements are separated. AEM elements are connected using the elements
entire surface, through a series of connecting springs as shown in Fig. 2.1. These
connecting springs represent stresses, strains and connectivity between elements. Table
2.1 shows the difference in element connectivity between FEM and AEM:
Table 2.1 Element connectivity comparison between AEM and FEM
FEM
AEM
Connectivity
Nodes (Joints)
Elements faces
Partial
connectivity
(during
analysis)
Not allowed
Allowed (Some springs may fail

while the others are still active
and effective)
Singularity
Separation of nodes causes stress Separation of springs does not

singularity
cause any type of singularity
Transition
Transition elements are needed

to switch from large sized
elements to smaller elements
No need for transition elements
Meshing
Very complicated specially at

interfaces with other objects
Every object is meshed alone

regardless of its connectivity to
surrounding elements
2.2.2.1 Transition from large-size elements to small elements

It is commonly needed that some zones of the analyzed structure are smaller in mesh-size
because of stress concentration and smaller crack spacing. In the FEM, Fig. 2.6, the
transition from large sized elements to smaller elements should be done through special
meshing techniques to assure connectivity between elements. This causes the following
problems:
1- Number of elements may increase due to transition layer
Complication in meshing process as elements should be connected through the boundary
(Refer to Fig. 2.5)
Transition elements
(a) FEM mesh
No transition elements needed
(b) AEM mesh

Fig. 2.6 Transition of element size
10
2.2.2.2 Partial Element Connectivity

In AEM, elements sharing the same surface will have connectivity springs even if the
shared surfaces are only a portion of the surface. Referring to Fig. 2.7, Finite Elements
are not connected at Fig. 2.7-a, while elements are connected with springs in the AEM as
shown in Fig. 2.7-b. In FEM, partial connectivity could actually be included, however,
more nodes should be then considered at the location of contacts, which means more
DOFs and more complication in element formulation.
No Connectivity
Connectivity included
(a) FEM
(b) AEM
Fig. 2.7 Partial element connectivity
2.3 Degrees of Freedom (DOF)

Referring to Fig. 2.8, each element has six degrees of freedom, three translations and
three rotations. The own weight always applies in the Z direction. The user can select the
active degrees of freedom in the problem to be solved as follows:
Table 2.2 Degrees of freedom for different analysis cases
Case
Sample
Active DOF
2-D Analysis in X-Z
2-D Multi story frame
Ux,Uz,Ry
Grid in X-Y
Floor, slab or Raft
Uz,Rx,Ry
General 3-D problem
Demolition of RC
Building
Ux,Uy,Uz,Rx,Ry,Rz
Uz
Uy
Ry
Rz
Ux
Rx
Fig. 2.8 Degrees of Freedom
11
It should be emphasized that the active degrees of freedom can have a significant effect
on the results. Here are some examples how a wrong selection of DOF may affect the
results:
Table 2.3 Possible DOF modeling errors
Case
Common mistake
Active DOF
2-D Buckling analysis
All DOF are active
Buckling may occur at out

of plan causing much less
buckling load
2-D Analysis
Supports are set in 2-D Plan

only while all DOF are
active
The problem is unstable
Grid Analysis under own

weight
All DOF are active
The problem may be

unstable
As DOF are the main unknown in the analysis, reducing the number of DOF will lead to
much faster analysis while the results are unaffected.
2.4 Supports
Structures are restrained against moment at support locations. These supports can be rigid
supports and/or elastic supports. Having rigid supports means that displacement should
be zero at the selected elements. Another way of introducing supports is to use Contact
Boundaries. These contact boundaries can exist at Minimum Z or Maximum Z.
At minimum Z connectivity springs are generated between the columns, or any attached
elements, and the contact boundary. Having columns supported by contact boundaries is
equivalent to having columns supported by elastic supports in both axial and rotations
DOF. To control stiffness of elastic support, the Contact Boundary has a specific
material ID representing the properties of connecting springs to the ground. Falling
elements collide with this boundary and rebound.
2.4.1 Rigid Supports

If the supporting elements totally prevent one or more displacement component, they are
called rigid supports. Having restraints at a specific joint leads to a reaction force or
moment due to the restrained degree of freedom. The following are some common rigid
supports.
2.4.1.1 Fixed supports
All displacements and rotational components are restrained; refer to Fig. 2.9. In that case
six components of reactions are calculated and all obtained displacements at that joint are
zero. A common example of a fixed support is the support at the column base connected
to a thick raft foundation.
There are two ways for fixing the end of the column shown in Fig. 2.9:
12
1. Select all the elements which contact the ground and fix them from displacement
(x, y, z), this method is enough to fix a column as shown in Fig (2.9)
2. Select one element in the center of the column base and fix it from displacement
(x, y, z) and Rotation about (x, y, z). Figure (2.10)
x
y
Fig. 2.9 Elements fixed from displacement in (x, y, z)
Z (Fixed)
Y (Fixed)
X (Fixed)
Fixed Element
Figure (1-1)
Fig. 2-10 An elements fixed from displacement and rotation in (x, y, z)
2.4.1.2 Hinged supports

All displacement components are restrained (zero displacement), while all rotational
components are free to rotate. As shown in fig. (2.11). In that case, reaction components
are forces calculated for restrained displacement components only. A common example
of hinged support is the support at a connection between a truss and its bridge piers.
13
For Hinged supports, select the elements shown in figure (2-11) and restrain them against
displacement in (x, y, z) and for stability restrain them against rotation around z axis.
2.4.1.3 Roller supports
In that case one or two displacement components are restrained (Fig. 2.11). That is the
roller is permitted to move in one or two direction. Reactions are calculated due to the
restrained direction only. A common example of a roller support is that of a truss with a
bridge pier. The user should be aware of the local joint axes directions when setting the
restraints of a roller support.
For roller supports, select the elements as shown in figure (2.11) and restrain them
against displacement in (y, z) and for stability restrain them against rotation around z
axis.
Z
Y
Hinged support
Roller support
Fig. 2.11 Hinged and roller supports
2.4.2 Planes Of Symmetry And Skew Symmetry

In some cases, structures are completely symmetric or skew-symmetric in geometry and
in loading, about one or more planes. In that case, it is wise to take advantage of
symmetry, especially when the structure is relatively large in size. It is worth noticing
that solving one-quarter of the structure shown in (Fig. 2.14), due to its symmetry about
X-Z and Y-Z planes, results in reducing the analysis time to around 1/50 the original
time. The main precaution in modeling half or quarter of the structure is to assign proper
boundary conditions. Figure 2.12 shows a sample of restrained degrees of freedom in
case of symmetry about the Y-Z plane. Figure 2.13 shows a sample of restrained degrees
of freedom in case of skew symmetry about the Y-Z plane. Figure 2.14 shows a sample
of restrained degrees of freedom in case of symmetry about both the X-Z plane & Y-Z
planes.
14
P
x
Fig. 2.12 Sample of restrained degrees of freedom in case of symmetry about the Y-Z plane
2P
z
x
y
Fig. 2.13 Sample of restrained degrees of freedom in case of skew symmetry about the Y-Z plane
P
4P
z
x
y
Fig. 2.14 Sample of restrained degrees of freedom in case of symmetry about the both X-Z plane &
Y-Z planes
15
2.4.3 Intersection Between Planes Of Symmetry

If planes of symmetry and/or skew symmetry intersect, joint restraints at the intersection
line should be set to the summation of restrained degrees of freedom of both planes. This
rule applies for intersection between two planes of symmetry, two planes of skew
symmetry and a plane of symmetry with a plane of skew symmetry.
2.5 Units
Basic units are used for the analysis. The user is advised to use a combination of massdistance-time unit. The time unit is always seconds. So, if the user selects the mass to be
in kilograms (Kg) and the distance to be in meters (m) then the forces will be in
(Kg.m/Sec2 / Gravity acceleration [m/Sec2]). In this case the force will be defined in
(Kilo Gram Force) or Kgf. Same if the user selects the mass to be in pounds and distance
to be in inches then the force unit will be (Pound.in/Sec2 / Gravity acceleration [in/Sec2]).
In this case it will be defined in Pound Force (lbf).
2.6 Material Properties

The mterial properties are defined for different structural elements used in Extreme
Loading for Structures (ELS). The materials here include concrete and reinforcement.
The following sections discuss the definitions of different material properties used.
2.6.1 Youngs Modulus

Youngs Modulus E is one of the main properties that define the behavior of an
element. It is defined as the ratio between longitudinal stress and strain in elastic range
for 1-D frame elements. Youngs Modulus cannot be equal to zero or a negative value.
The deformations of the structures in the elastic stage depend very much on the young's
modulus. For example, assuming that all structural elements (columns, beams, slabs)
composed of only one type of material (like steel structures); Increasing the Youngs
Modulus to double its value, leads to the following:
1- Generated displacements are half their original values.
2- All obtained internal forces (bending moments, shear forces, etc.) and stresses in
shell elements are the same without any changes.
3- All obtained rigid reactions are the same.
Hence, setting a precise value for Youngs Modulus is important, only in the following
cases:
1- Structure is composed of different materials, like RC structures are composed of
steel and concrete elements.
2- Precise displacement values are required.
16
Stress
Stress
E
Strain
Strain
Steel
Concrete
Fig. 2.15 Young's modulus
The units of Youngs modulus should be force/Distance2

Mass.Distance/Sec2/Distance2/Gravity Acceleration [Distance/Sec2]. For example,
using "meters" for distance and "kg" for mass, the Youngs Modulus should be in
"kgf/m2". If the distance is in "inches" and mass to be in pounds then the Youngs
Modulus will be lbf/in2.
2.6.2 Shear Modulus

Shear Modulus G is the ratio between shear stress and shear strain within the elastic
range for a 2-D element. Its value is defined as:
G=
E
2(1 + )
(2-1)
Where E is the Youngs Modulus and is the Poissons ratio. The shear modulus
value is usually around 0.4 E and hence, cannot be zero or negative.
The units of Shear modulus are force/Distance2 Mass.Distance/Sec2/Distance2/Gravity
Acceleration [Distance/Sec2]. For example, using "meters" for distance and "kg" for
mass, the shear Modulus will be in "kgf/m2". If the distance in "inches" and mass to be in
pounds then the shear Modulus will be lbf/in2.
17
Shear Stress
G
Shear Strain
Fig. 2.16 Shear modulus
2.6.3 Tensile Strength (Or Tensile Yield for Steel)

The tensile strength is the stress at which rupture takes place. For steel, tensile strength
here represents the tensile yield stress as follows in Table 2.4;
Table 2.4 Values of tensile strength for different materials
Material
Description
Concrete
This value represents the tensile strength of concrete (Fig. 2.17-a)
Steel
This value represents the yield strength of steel (Fig. 2.17-b)
Elastic
No effect on the results
The units of Tensile strength are force/Distance2 Mass.Distance/Sec2/Distance2/Gravity

Acceleration [Distance/Sec2]. For example, using "meters" for distance and "kg" for
mass, the tensile strength will be in "kgf/m2". If the distance in "inches" and mass to be in
pounds then the tensile strength will be lbf/in2.
Stress (+)
Stress (+)
ft (yield)
ft
Strain (+)
(a) Concrete
Strain (+)
(b) Steel
Fig. 2.17 Tensile strength
18
2.6.4 Compressive Strength (Or Compressive Yield for Steel)

The compressive strength is the stress at which compression failure takes place. For steel,
compressive strength here represents the compressive yield stress as follows in Table 2.5;
Table 2.5 Values of compressive strength for different materials
Material
Description
Concrete
This value represents the compressive strength of concrete (Fig.

2.18-a)
Steel
This value represents the compressive yield strength of steel which

is usually same value as the tensile yield (Fig. 2.18-b)
Elastic

Stress (-)
Stress (-)
fc (yield)
fc
Strain (-)
(a) Concrete
Strain (-)
(b) Steel
Fig. 2.18 Compressive strength
The units of compressive strength are force/Distance2

Mass.Distance/Sec2/Distance2/Gravity Acceleration [Distance/Sec2]. For example,
using "meters" for distance and "kg" for mass, the compressive strength will be in
"kgf/m2". If the distance in "inches" and mass to be in pounds then the compressive
strength will be lbf/in2.
2.6.5 Ultimate strength/ yield stress

The ultimate-strength/ yield stress is the ratio of the ultimate strength of reinforcing bars
to its yield stress.
19
Table 2.6 Ultimate-strength/ yield stress
Material
Description
Concrete
Steel
Ratio between Ultimate strength and Yield Strength (Fig. 2.19).

Default value is 1.3
Elastic
The Cut-off-ratio has no units. In case the reinforcement bars are not required to cut, a
large value may be set as a cut off ratio.
Stress
Ultimate
strength
Yield stress
Strain
Fig. 2.19 Ultimate strength/yield stress for reinforcement
2.6.6 Separation Strain

Separation strain represents the strains at which adjacent elements are totally separated at
the connecting face as shown in Fig. 2.20.
= Separation
Strain
Fig. 2.20 Separation strain
20
Table 2.7 Separation strain for different materials
Material
Description
Concrete
Strain at which elements are separated. All springs between the

adjacent faces including reinforcement bar springs are cut. If the
elements meet again, they will behave as two different rigid bodies
that contacted (Fig. 2.21).
Steel
No effect on the results. The steel bars are cut if its stress reaches
the ultimate stress or if the Concrete reaches the separation strain
Elastic
The Separation strain has no units. The default value is 0.1. It is recommended to use
value between 0.1 and 0.2.
Initial position
Element 1
Element 2
Matrix springs under tension

Element 1
Element 2
Separation strain reached

Element 1
Element 1
Element 2
Element 2
Element 1 Element 2
Elements forced towards each

other (according to loading
condition)
Contact springs activated
Fig. 2.21 Elements separate and re-contact again
2.6.7 Friction Coefficient

The Friction Coefficient value is not effective before the elements are cut. It affects only
after element separation when elements meet as rigid bodies. The Friction Coefficient has
no units.
21

Table 2.8 Friction coefficients
Material
Description
Concrete
Steel
This value affects only when contact occurs between elements and
has no effect if no contact occurs (Fig. 2.22)
Elastic
Fig. 2.22 Friction between two contacting elements
2.6.8 Specific Mass

The specific mass value is the mass per unit volume. It has units of (mass/Distance3). The
Specific gravity is calculated automatically in the program as follows:
Sp. Gravity = Sp. Mass x Gravity Acceleration
(2-2)
It should be noted that having a zero value of specific mass will lead to:
1. Own weight is not considered. Falling objects will not fall down.
2. All dynamic analysis features will be disabled.
The specific mass value is used to calculate the element weight. It has no effect if the
material is referring to reinforcement bars as reinforcement bar mass and weight is
neglected. But when solving a steel structure, this value is used to get the element weight.
2.6.9 Damping Ratio r

Damping is always classified as external damping and internal damping. The internal
damping is caused by many factors including:
1. Cracking of concrete
2. Unloading of concrete or steel bars after yield
3. Friction between elements during contact
4. Unloading factor when contact occurs, as will be shown later
The above factors are main damping parameters. But they are effective only when
nonlinear analysis is performed. Assuming the material is elastic, all above factors will
not occur. In that case a damping factor should be used to assure reasonable energy
dissipation.
Assuming the general dynamic equation:
22
m&u& + cu& + ku = f (t )
(2-3)
The damping factor r is the ratio between (c/m). To have damping ratio of 0.05,
then the user should do the followings:
1. Estimate a reasonable natural period T
2. Calculate the Natural frequency W = 2Pi/T
3. Estimate the Damping Factor = 2 * * W
This value has a default value of zero as in most of nonlinear analysis cases, internal
damping is sufficient for good accuracy.
It should be emphasized that having external damping causes elements not to fall by
gravity acceleration as damping acts as deceleration forces for falling objects.
2.6.10
Post-Yield Stiffness Ratio
Referring to Fig. 2.23, the post-yield stiffness ratio is the ratio between post-yield and
pre-yield tangent modulus of steel. It has an effect for only springs with steel properties;
either reinforcement bars or in cases where matrix material itself is steel, as with steel
beams.
Stress
E1
fy
E
Strain
Fig. 2.23 Post-yield stiffness ratio
Table 2.9 Post-yield stiffness ratio
Material
Description
Concrete
Steel
Ratio between post-yield stress-strain relation and before yield.

This ratio has a default value of 0.01
Elastic
23
2.6.11
Contact Parameters
When contact occurs between elements, springs are generated at contact points. Those
springs are linear springs that transfer energy between elements. Contact spring stiffness
should have a reasonable value. Having very high contact stiffness causes a large shock
force to transfer between elements and also having low contact stiffness causes the
underestimation of transmitted forces when elements collide. So, reasonable stiffness
value should be assumed when elements collide. The following defines the main
parameters.
2.6.11.1 Normal Contact Stiffness Factor (NF)
Referring to Fig. 2.24, for the two elements in contact. The normal spring stiffness is
calculated as:
Stiffness = E * A/ D
(2-4)
Where;
"E" is the minimum Youngs modulus of the two elements
"A" is the average contact area between elements
D is the center to center distance between the two elements (assumed in the analysis as a
constant value calculated from average element size). So, the user does not have control
over the value of D
Contact area (A)
D
NF*k
Fig. 2.24 Normal contact stiffness factor
Assuming that D is a ratio of the Average face area, A = NF * D2

Finally, the spring stiffness is calculated as E * NF * D. The value of NF is an input
parameter named Normal Stiffness Factor
The above equation is used to calculate the stiffness value of normal spring when loading
occur. Default value of Normal Stiffness Factor is 0.01
2.6.11.2 Shear Contact Stiffness Factor (SF)
Shear Stiffness factor is very similar to the concept of Normal Stiffness Factor. However,
as transmitted normal force is always higher than the transmitted shear force, its default
value = 0.001
24
The shear spring stiffness is affected mainly by friction properties. When shear force is
less than friction coefficient x Normal force, then the shear stiffness = G * SF * D.
Otherwise, the shear stiffness is G * SF * D / 1000.0 because elements are supposed to be
sliding over each other.
Contact
area (A)
SF*k/1000
D
SF*k
Fig. 2.25 Shear contact stiffness factor
2.6.11.3 Contact Spring Unloading Stiffness Factor

When elements collide some energy is dissipated during contact. Referring to Fig. 2.26,
ratio between loading and unloading stiffness is called n. This ratio controls how much
energy is dissipated during contact. Having n = 1 means that no energy dissipated
during contact and element will rebound by the same speed before colliding (Fig. 2.27).
When n > 1 then the hatched area represents the energy dissipated during contact. Default
value of n = 10 which means that the energy dissipated is 90% of the element energy is
dissipated during contact and only 10% is recovered. This value is decided by
engineering sense of the user. The value of 10 is reasonable if the user does not have an
estimation of a more accurate value.
b
a
k
a
nk
c
Fig. 2.26 Contact spring unloading stiffness factor
25
n=1
n=10
Fig. 2.27 Rebound with different values of contact spring unloading stiffness factor
2.7 Connectivity Springs

AEM elements are connected together through a series of springs that connect adjacent
element faces. The generation of these springs is automatically performed in ELS. These
springs are very important because:
1. They represent continuity between elements. The user is strongly advised to check
the solved model using the Viewer and Spring View option to check whether
springs are generated or not
2. These springs have all material properties. Strains, stresses, reinforcement bars
and failure criteria are all calculated and estimated at these springs
2.7.1 Matrix Springs

Referring to Fig. 2.28, Matrix springs are those springs that connect two adjacent
elements and are representing the main structural material. For example, when solving a
reinforced concrete structure, these springs represent the concrete part. When solving
steel structure, these springs represent the steel part and so on. These springs adopt all
material type and properties defined in the GUI. When the average strain between these
two adjacent faces reaches a specific limit, which is defined in the material property
section in the GUI, springs between these two faces are removed and it is assumed that
these elements behave as rigid body for the remainder of the analysis. Even if elements
meet again, they meet as a contact between rigid bodies. At every calculation point, three
springs are set for matrix springs. One for normal stresses and the other two springs are
for the shear springs. Average normal strain is calculated by having the average of
absolute values of strains at each face. When the average strain value at the element face
reaches the separation strain, all springs at this face are removed and elements are not
connected any more until they collide. If they collide together they collide as rigid bodies.
The user should be aware that to generate the springs between faces, the faces should be
exactly in the same plan. Hence the user is strongly requested to use Snap options in the
26
GUI to draw the problem to be solved so that all faces are exactly in the same plan with
minimum or no tolerance. Otherwise, springs may not be generated at connecting faces,
as shown in Fig. 2.29.
Element 1
Element 2
Element 1
Normal Springs
Element 1
Element 1
Shear Springs x-z
Shear Springs y-z
Fig. 2.28 Connectivity matrix springs

Connectivity
springs generated
No connectivity
springs generated
>>
Fig. 2.29 Precautions required to assure creation of connectivity matrix
2.7.2 Reinforcement Springs

In reinforced concrete structures, the Matrix Springs are those springs representing the
concrete material. The reinforcement springs are those springs representing steel bars as
shown in Fig. 2.30
27
Element 1
Element 2
Reinforcing
bar
Element 1
Normal Springs
Element 1
Shear Springs x-z
Element 1
Shear Springs y-z
Fig. 2.30 Reinforcement springs
The reinforcement springs have material properties, exact location and dimension of the
represented reinforcement bar. Similar to matrix springs, three springs are set at the
intersection of the reinforcement bar and the element boundary, as shown in Fig. 2.30.
The normal spring takes the direction of reinforcement bar irrespective to the element
face direction. The other two springs represent reinforcement bar behavior in shear. The
reinforcement bar springs are cut in one of the following cases:
1. The reinforcement bar stresses reaches the failure criteria. The failure criterion is
reached by having a normal stress equal or greater than the ultimate stress. It
should be emphasized that the bars rupture works only with bars in tension. No
cut is permitted for steel bars having compression forces.
2. Or, the matrix springs reach the Separation Strain limit. In that case all matrix
springs and reinforcement springs either in tension or compression are removed.
2.8 Analysis
The overall equilibrium set of equations in the dynamic problem is as follows;
[M]{X}oo + [C]{X}o + [K ]{X} = {f }
(2-5)
The solution for dynamic problems adopt the step-by-step integration (Newmark-beta)
method. The equilibrium equations are actually a linear system of equations for each step.
The solution of the equilibrium equations in ELS is commonly solved using either a
direct solver (Cholesky upper-lower decomposition) or an iterative solver. The iterative
solvers are in general faster than direct ones especially for large numbers of degrees of
freedom.
28
2.8.1 Direct Solver

2.8.1.1 Choleski factorization with upper-lower decomposition
For solvers that use the direct method, the basic technique employed in finding the
solution of the system [K]{x} = {b} is to first factor (decompose) [K] into triangular
matrices. That is, find a lower triangular matrix [L] and an upper triangular matrix [U],
such that [K] = [L][U]. Having obtained such a factorization (usually referred to as an LU
decomposition), the solution to the original problem can be rewritten as follows.
[K]{x} = {b}
(2-6)
[L] [U] {x} = {b}
(2-7)
[L]([U] {x}) = {b}
(2-8)
This leads to the following two-step process for finding the solution to the original
system of equations:
1. Solve the systems of equations [L] {y} = {b}
2. Solve the system [U] {x} = {y}
Solving these two systems is referred to as a forward solve and a backward solve,
respectively.
If a symmetric matrix, [K], is also positive definite, it can be shown that [K] can be
factored as [L][L]t where [L]is a lower triangular matrix. The factorization of this form is
called a Cholesky factorization.
In a Cholesky factorization, the matrix [U] in an [L][U] decomposition equals [L]t.
Consequently, a sparse solver can increase its efficiency by only storing [L], and one half
of [K], and not computing [U].
2.8.1.2 CXML direct solver 12)
The stiffness matrix [K] of typical structural systems has only very few non-zero entries.
Since most elements of an actual problem discretization are only neighbors to few other
elements compared with the number of elements of the whole system, the population of
K with non-zero entries is very sparse.
Modifying a standard Cholesky decomposition, in a way that, it only operates on the nonzero elements of the matrix [K] leads to a faster solver (Compaq Extended Math Library
(CXML), 200112)). Furthermore one has to store only the nonzero elements within the
stiffness matrix, reducing the memory needs drastically.
As mentioned above, it is more efficient to store only the non-zeros of a sparse matrix.
There are a number of common storage schemes used for sparse matrices, but most of the
schemes employ the same basic technique. That is, compress all of the non-zero elements
of the matrix into a linear array, and then provide some number of auxiliary arrays to
describe the locations of the non-zeros in the original matrix.
29
The compression of the non-zeros of a sparse matrix into a linear array is done by
walking down each column (column major format) or across each row (row major
format) in order, and writing the non-zero elements to a linear array in the order that they
appear in the walk. This technique (CXML12) direct sparse solver) reduces storage
capacity even more than the sky-line solvers which store the rows (or columns) up to the
last non-zero element. In fact the sky-line solvers store the zero elements lying between
the matrix diagonal and the last non-zero element in each row or column (see Fig. 2.31).
When storing symmetric matrices, it is necessary to store only the upper triangular half of
the matrix (upper triangular format) or the lower triangular half of the matrix (lower
triangular format). The CXML12) direct sparse solver uses a row major upper triangular
storage format. That is, the matrix is compressed row-by-row and for symmetric matrices
only non-zeros in the upper triangular half of the matrix are stored. In CXML12), direct
sparse solvers cannot solve non-symmetric systems of equations. However, it can solve
symmetrically structured systems of equations. A symmetrically structured system of
equations is one where the pattern of non-zeros is symmetric. That is, a matrix has a
symmetric structure if K(i; j ) is non-zero if and only if K(j; i) is non-zero.
Two important concepts associated with the direct solution of sparse systems of equations
are fill-in and reordering. The fill-in concept means that the [L] matrix might not be as
sparse as the [K] matrix. In other words, the [L] matrix might have no or few zeros below
the diagonal, which is not favorite in matrices solutions. Consequently, if we computed
[L] and then used it for the forward and backward solve phase, we would do as much
computation as if [K] had been dense. Computationally, it would be more efficient if a
solver could exploit the non-zero structure of [K] in such a way as to reduce the fill-in
when computing [L]. By doing this, the solver would only need to compute the non-zero
entries in [L]. Toward this end, the CXML12) direct solver considers the permuting the
rows and columns of the matrix [K]. This leads to finally getting a non-dense [L] matrix.
Direct method allows for moderately large systems to be solved efficiently. However still
found to be unsuitable for solving very large systems, especially those for 3- dimensional
problems with large number of degrees of freedom. Moreover, for progressive collapse
problems, the solution sometimes depends entirely on the mass matrix rather than the
stiffness matrix, which means that the direct method can not be used.
30
CXML Solver
SkySky-Line Solver
Sparse Matrix
Store elements up the last non

zero element in each row
Store only the nonzero elements

reducing the memory needs drastically
Fig. 2.31 Reduction of storage capacity in CXML solver 12)
2.8.2 Iterative Solver

The iterative solver is a more practical method than a direct solver for solving large
systems of equations, since they require less CPU time and storage space for comparable
accuracy. This is because the global matrix need not be formed and the major operation is
a multiplication of a matrix and a vector (as opposed to matrix reduction which is done in
direct methods). The performance of the iterative solver most crucially depends on
elective preconditioning and fast matrix multiplication.
Iterative solvers start with an initial approximation to a solution and attempt to estimate
the difference between the approximation and the true result. Based on the difference, an
iterative solver calculates a new approximation that is closer to the true result than the
initial approximation. This process is repeated until the difference between the
approximation and the true result is sufficiently small. The main drawback to iterative
solvers is that the rate of convergence depends greatly on the values in the matrix [K].
Consequently, it is not possible to predict how long it will take for an iterative solver to
produce a solution. In fact, for ill-conditioned matrices, the iterative process will not
converge to a solution at all. However, for well-conditioned matrices it is possible for
iterative solvers Using Direct Sparse Solvers to converge to a solution very quickly.
Consequently for the right applications, iterative solvers can be very efficient.
The direct solver is generally optimum for static analysis or dynamic analysis with
relatively large time increment, while the iterative solver is optimum with dynamic
analysis with very small time step. The ELS tests both solvers and switches between
them automatically. Whenever the user switches between static stage and dynamic stage,
the optimum solver is tested and allocated.
31
3 Material Models and Failure Criteria

Each reinforcement spring or matrix spring is assigned a specific material model. In the
current version, three material models exist as follows.
3.1 Elastic Model

When a material is assumed elastic, the stress-strain ratio is continuously constant. This
ratio is the Youngs Modulus. Stresses and strains can reach very high values without any
changes in the material behavior.
3.2 Reinforced Concrete Models

3.2.1 Concrete models
Figure 3.1 shows the constitutive models adopted in ELS. As for modeling of concrete
under compression, Maekawa compression model9), as shown in Fig. 3.1-a, is adopted. In
this model, the initial Young's modulus, the fracture parameter, representing the extent of
the internal damage of concrete and the compressive plastic strain are introduced to
define the envelope for compressive stresses and compressive strains. Therefore
unloading and reloading can be conveniently described. The tangent modulus is
calculated according to the strain at the spring location.
After peak stresses, spring stiffness is assumed as a minimum value to avoid negative
stiffness. This results in difference between calculated stress and stress corresponds to the
spring strain. These residual stresses are redistributed by applying the redistributed force
values in the reverse direction in the next loading step.
For concrete springs subjected to tension, spring stiffness is assumed as the initial
stiffness until reaching the cracking point. After cracking, stiffness of springs subjected to
tension is set to be zero. The residual stresses are then redistributed in the next loading
step by applying the redistributed force values in the reverse direction.
For concrete springs, the relationship between shear stress and shear strain is assumed to
remain linear till the cracking of concrete. Then, the shear stresses drop down as shown
in Fig. 3.1-b. The level of drop of shear stresses depends on the aggregate interlock and
friction at the crack surface.
3.2.2 Reinforcing bars Model

For reinforcement springs, the model presented by Ristic et al.10) is used and it is shown
in Fig. 3.1-c. The tangent stiffness of reinforcement is calculated based on the strain from
the reinforcement spring, loading status (either loading or unloading) and the previous
history of steel spring which controls the Bauschinger's effect. The main advantage of
this model is that it can consider easily the effects of partial unloading and Baushinger's
effect without any additional complications to the analysis. The rupture strain of
reinforcement is defined in ELS.. For more details about material models used, refer to
Tagel-Din 4).
32
3.2.3 Cracking Criterion Of Concrete

The concrete is assumed to crack when the major principal stress reaches the tensile
strength of concrete. The 3D state of stresses at each spring location and the major
principal stress is calculated as shown in Fig. 3.2 After cracking, there are mainly two
ways to consider the crack if the crack is not coinciding on the element surface;
1- to split the element into two elements and to generate new springs among the
crack surface
2- to leave the element as it is and to redistribute stresses resulting from cracking
The first method is generally more accurate but it is very complicated and time
consuming especially when computing large problems and progressive collapse cases.
The other method is not accurate but it gives reasonable results. If the shear cracks
govern the behavior, it is advisable to reduce the element size in order to get cracks close
to reality.
Friction and
interlocking
(a) concrete under axial stresses
(b) concrete under shear stresses
(c) Reinforcement under axial stresses

Fig. 3.1 Constitutive models for concrete and reinforcement used in ELS
33

z
xz
x
yz
xy
zy
yx
x
zx
yz xz x
y
xy
zy
yx
zx z
x
Plane of major principal stress
Fig. 3.2 Cracking criterion of concrete
Added springs
at fracture plan
Splitting Elements
Redistributing stresses at element edge
(Complicated but accurate)
(Simple but not accurate for shear

transfer problems)
Fig. 3.3 Behavior of concrete elements after cracking
3.3 Composite Steel-Concrete Models

A composite member is modeled as shown in Fig. 3.4. The cross section is divided into
cells where steel springs are put at the location of steel section elements. The default
interface material between concrete and steel is concrete. Here, it should be mentioned
that there are two main assumptions for the modeling of composite sections;
34
1- The interaction between the normal stresses in the two perpendicular directions
of the plates constituting the steel section is not taken into consideration. This is
in fact acceptable when dealing with frame structures where the longitudinal
direction represents the main direction of stresses in the steel section.
2- The connections between any two members are assumed not to fail. The possible
failure in connections is then not predictable by the ELS
As for bare steel members, the same concept is followed where only elements containing
parts of the steel section are used while other elements are considered as vacuum
elements with no material inside as shown in Fig.3.5.
Springs in section
Concrete Cells
Steel Springs
S
di prin
re g
ct s
io in
n
lo
ng
itu
di
na
l
The steel behavior follows the same models used above for reinforcing bars, except for
the failure criterion, where the failure criterion is based on the principal tensile strain as
with the concrete case.
Concrete Springs
Fig. 3.4 Modeling of Composite Steel-Concrete Members
35
Springs in section
Vacuum cells
S
di prin
re g
ct s
io in
n
lo
ng
itu
di
na
l
Steel Springs
Fig. 3.5 Modeling of Steel Members
3.4 Masonry Wall Models

Masonry walls can be simulated in ELS either alone or inside reinforced concrete or steel
frames as shown in Fig.3.6. Windows can also be simulated in any arbitrary location
inside the wall. The window is composed of glass with outer and inner aluminum frames
as shown in Fig.3.6.
The brick can be simulated either in a staggered pattern, in its real distribution, or as a
continuum like concrete as shown in Fig. 3.7. The real distribution includes the individual
bricks in a staggered pattern connected by mortar. However, it is possible to divide the
brick itself into sub-elements to allow the brick to be broken from its inside. The
continuum simulation, or the macro simulation, represents the brick wall as one
homogenous material. This model is also reasonable and acceptable especially when
analyzing concrete masonry units.
The behavior of the bricks and the mortar are assumed to be similar to concrete's
behavior mentioned above. It should be mentioned that, the possible shear-compression
failure due to highly applied axial loads on walls is not taken into consideration in the
current version of ELS.
The aluminum behavior is very similar to the steel's one mentioned above with different
Young's modulus, yield stress and ultimate strength.
The glass is a brittle material. It follows a linear elastic behavior up to tension failure.
The fracture process and fragmentation of glass occurs randomly and needs special care
36
in meshing. Therefore, and in order to obtain a realistic reasonable fracture pattern of

glass windows in ELS, the so called "Voronoi Diagram" or "Delaunay Triangulation" is
adopted as shown in Fig. 3.8.
Interface
Aluminum
Mortar
Brick
Mortar
Concrete
Glass
Fig. 3.6 Modeling of Masonry Walls
37
Real simulation
Macro simulation
Brick + mortar
One homogenous material
Fig. 3.7 Different Meshing for Masonry
Fig. 3.8 Delaunay Triangulation "Voronoi Diagram" Meshing in ELS
38
4 Loading
ELS software contains many types of loadings. These loadings are:
1- Own weight
2- Static loading
a. Load control
b. Displacement control
3- Dynamic loading
a. Dynamic load control
b. Dynamic displacement control
c. Earthquakes
d. Blast loads
4.1 What are loading stages?

To understand what is meant by loading stages, have a look at the following table:
Table 4.1 Loading stages in ELS
Case
Linear Analysis
Nonlinear Analysis
Dead + Live
loads
Both loads can be applied at the same

time or the results of both analysis are
added together
Sequence of loading is important, own weight and

dead loads should be added first and then live loads
Earthquakes
Results from static analysis for own

weight can be added to the results of
dynamic analysis under earthquake to get
final results
Static loads should be applied first to the structure

then the earthquake. The earthquake is applied to the
structure with initial stresses resulting from gravity
loads.
A projectile
impacting
a
building
Can not be performed
In the first stage, the projectile is fixed and the

building is solved under own weight and static loads
In the second stage, the projectile is a) released from
fixation, b) given initial velocities in all directions to
hit the building. In this stage, the building can be
fixed optional to save time and memory
If the building is fixed in the second stage then we
need a third stage that is activated when the rock
approaches the building. The user should judge
appropriate time for the second stage.
In the third stage, when the rock approaches the
structure, the building is released and collapse
analysis for the projectile and building is performed.
From the above table, it is critical to apply loads at separate stages. In each stage the user
is allowed to:
1- Define the stage loading type, static or dynamic
39
2- Boundary conditions, or supports, can be variable with each stage.

3- Initial velocity can be defined for a group of elements at each stage.
40
5 Loading Types
ELS performs analysis for both static and dynamic loading scenarios. Each loading stage
must be defined separately as either static or dynamic.
5.1 Static Loading

5.1.1 Own Weight
Own weight loading is the consideration of the weight of structural elements themselves.
The first stage of loading is allocated for considering the own weight which by default
exists in all subsequent loading stage. Hence the first loading stage should be defined as
Static or own weight stage. In some cases, special considerations should be made to
include the own weight effect as:
1- In cases where a falling object collides with a building. The overall behavior is
unstable for the own weight loading stage due to the existence of an
unsupported object in the air. This case can be solved in dynamics only but can
not be solved in static. Hence, to include the own weight of the structure, the
falling object should be fixed in the first stage then solve the whole structure
under own weight. In the second stage, solve the whole system in dynamic while
the falling rock is released.
2- In case of a pendulum to be solved, the pendulum is originally unstable. Hence, it
can not be solved under own weight in static. So, in the first stage, fix all elements
of the pendulum and release them in the second stage dynamic to start the
motion.
5.1.2 Lumped Weight (mass)

The own weight feature accounts for weights of elements to be solved. The nonstructural elements weight can be accounted for by:
1- Increasing the specific weight of the elements to account for other weights. For
example, the reinforced concrete specific gravity is usually 2.5 ton/m3. To account
for other weights, live loads and flooring, a new specific weight value should be
calculated so that the other weights are considered. Suppose that a slab with
thickness 0.15 m with a specific gravity of 2.5 ton/m3, Live loads = 0.2 ton/m2
and flooring = 0.15 ton/m2. Hence the total weight per square meter = (0.15 x 2.5
+ 0.2 + 0.15) = 0.725 ton/m2. The new specific gravity for concrete should be
0.725/0.15 = 4.833 ton/m3 instead of 2.5 ton/m3.
2- The above technique works well with distributed weights and masses. But in
some cases the load is concentrated in one point or line. Hence, additional weights
should be added to the elements at these locations to account for the wall loads
using lumped mass.
3- Lumped mass/weight works also fine to add additional weights for the whole
floor elements as an alternative for technique #1.
41
5.1.3 Load Control

Static load control is used to add concentrated loads to specific elements. These loads can
be in the form of loads in X, Y and Z or moments around X, Y and Z, respectively. The
applied static load is applied in equal increments calculated by dividing the applied load
by number of increments.
In static load control, the ultimate load is predefined. Fig. 5.1 shows the parameters that
should be entered to define the loading stage for load control case. This ultimate load is
reached through a defined number of loading increments. The more the number of
loading increments are, the more the solution accuracy and convergence. In static
loading, the output time step is usually assigned to unity. In the initial conditions, six
values are available which represent the six components of ultimate load.
Fig. 5.1 Parameters defining load control
5.1.4 Displacement Control

Static displacement control is used to add concentrated loads to specific elements. These
displacement loads can be in the form of predefined displacements in X, Y and Z or
predefined rotations around X, Y and Z, respectively. The applied static load is applied in
equal increments calculated by dividing the applied load by number of increments.
The main difference between displacement control and load control is that in
displacement control the displacement is known while the applied load is unknown. On
the other hand, in load control cases, the applied load value is known while the resultant
displacement is unknown.
One main advantage of displacement control is that it can follow the post-peak behavior
so failure load can be calculated easily. But in case of load control, usually the solver
stops when reaching the failure load.
The main disadvantages of displacement control are that:
1- It can not be used for distributed loads as the load value is not known
2- It is not recommended in cases of more than one loading location
In case the same problem is to have both concentrated loads together with displacement
loads, they can not be applied together in the same stage. So, you must define more than
one loading stage and put the displacement loads in one stage and the concentrated loads
in another one.
In static displacement control, the ultimate displacement is predefined. Fig. 5.2 shows the
parameters that should be entered to define the loading stage for displacement control
42
case. This ultimate displacement is reached through a defined number of displacement

increments. The more number of displacement increments are, the more the solution
accuracy and convergence. In static displacement, the output time step is usually assigned
to be as one. In the initial conditions, six values are available which represent the six
components of ultimate displacement.
Fig. 5.2 Parameters defining displacement control
5.2 Dynamic Loading

Dynamic loads are the loads that vary with time. The structural behavior, internal force,
geometry is also a function of time. If the user decides to work with dynamic loads, then:
1- The time step for analysis should be selected soundly. If the behavior includes
collision, then the time step should be small when collision occurs.
2- If the user sets the element mass as zero, then the analysis is static as the inertia
forces will be zero.
5.2.1 Calculation Time Step

Time effects are continuous since the start of loading till the end of analysis. As the
nonlinear dynamic phenomenon is very complicated to be solved using exact solutions,
approximate numerical solution for the dynamic equation is adopted. The numerical
solution is based on assumption of a small time step that can follow the structural
behavior
The selection of the time step is very important. Having too short a time step will result in
very long analysis time. Using a large time step thus a short analysis time will result in
less accurate analysis. The numerical solution may fail to converge if the selected time
step is large. Say that the time step is T, then the shortest period that can be considered
in the analysis is 2 T (Highest frequency is Pi/ T). All frequencies higher than thus
frequency will not affect the analysis.
The applied excitation is applied as a record (in earthquake excitation), with a recording
time step. For calculation purposes, the recording time step is subdivided into smaller
steps called calculation time steps (i.e. calculation time step is smaller than the recording
time step). The ratio of recording/calculation time step is usually bigger than unity. The
bigger the ratio is the better the accuracy and the convergence of results. Earthquakes
usually required t of 0.001-0.01 Sec. for analysis. When collision is expected to occur,
the time step should be less than the contact time between elements to have good contact
relations. Referring to Fig. 5.3, in ELS, The ratio of recording/calculation time step is
entered as the number of intervals before and after collisions between impacting elements
in motion.
43
5.2.2 Load Control

In dynamic load control, the time series for applied load excitation is user-defined. Fig.
5.3 shows the parameters that should be entered to define the loading stage for dynamic
load control case. Note that the load is applied as a (Stage 2) after the own weight is
applied in (Stage 1). Loads are applied in accordance with the original delta time entered.
The maximum calculation time entered determines the total number of rows in the
loading table. Intervals before and after collisions represent the subdivisions of the
recording load time step. The results can be output at the every calculation time step or at
multiples of the original delta time. To enter the load series, you can directly enter the
values in the table shown in Fig. 5.3. In the fill data block, six values are available which
represent the six components of ultimate load. Another method for entering the load
series is to import the series from a saved text file. You must first determine to which
component would the read date saved. The data read can be scaled by multiplying them
by any factor. Data must be organized in the text file in one column with the first row
representing the total number of rows to be imported and subsequent rows hold the values
of loads to be imported for every delta time.
Fig. 5.3 Parameters defining dynamic load control
5.2.3 Displacement Control

In dynamic displacement control, the time series for applied displacement excitation is
user-defined. Fig. 5.4 shows the parameters that should be entered to define the loading
stage for dynamic displacement control case. Note that the displacement is applied as a
(Stage 2) after the own weight is applied in (Stage 1). Displacement is performed on the
selected model elements at increments of the original delta time entered. The maximum
calculation time entered determines the total number of rows in the loading table..
Intervals before and after collisions represent the subdivisions of the recording
44
displacement time step. The results can be output at the every calculation time step or at
multiples of the original delta time. To enter the displacement series, you can directly
enter the values in the table shown in Fig. 5.4. In the fill data block, six values are
available which represent the six components of ultimate displacement. Another method
for entering the displacement series is to import the series from a saved text file. Data
must be organized in the text file in one column with the first row representing the total
number of rows to be imported and subsequent rows hold the values of loads to be
imported for every delta time. The data read can be scaled by multiplying them by any
factor.
Fig. 5.4 Parameters defining dynamic displacement control
45
Fig. 5.5 Parameters defining earthquake loads
5.2.4 Earthquake
In earthquake dynamic analysis, the time series for applied earthquake acceleration
excitation is user-defined. Fig. 5.5 shows the parameters that should be entered to define
the loading stage for earthquake input wave. Note that the seismic accelerations are
applied following the own weight application stage (Stage 1). Earthquake loads are
applied at steps based on the delta time. The maximum calculation time is the total time
for applying earthquake loads. It determines the number of delta time steps for entering
loads. Intervals before and after collisions are the computational increments that
determine the increments for the solver to perform calculations. The results can be
output at the every calculation time step or at multiples of the original delta time. To
enter the earthquake load series, you can directly enter the values to the table shown in
Fig. 5.5. In the seismic acceleration data block, shown in Fig. 5.5, three values are
available which represent the three components of seismic acceleration. Another method
for entering the displacement series is to import the series from a saved data file "pressing
Import Acceleration Data". You must first determine to which component would the read
data be saved. The data read can be scaled by multiplying them by any factor. Multiply
text box is used for magnification of the imported acceleration..
5.2.5 Inertia Loads from Lumped Mass

Inertia forces are induced in dynamic analysis. Inertia forces are the multiplication of
acceleration by mass. The mass can be distributed or lumped at a small volume. A
lumped mass can be assigned to any group of elements using the lumped Mass"
window. Simply click Select Elements to choose the elements that will carry the assigned
mass. Enter the Mass Value in the designated field and click Assign to Selected Elements
to assign the mass to the selected elements.
46
5.2.6 Blast Scenario

Blast effects are modeled using free-field models of blast waves. The pressure resulting
from the blast wave is a function of bomb weight, distance to the bomb and time. The
free field model is used in two alternate methods. The first model employs an empirical
equation11). The second method utilizes discrete data obtained experimentally. Friedlander
equation11) is used to represent the pressure-time history at any point of the structure. The
bomb pressure forces are calculated as follows:
1- For each element face, it is determined whether the ray extending from the
element face to the bombing source intersects with other structural elements or
not.
2- If no intersection, this element is considered loaded by direct pressure wave.
Other elements are assumed not to be affected directly by bomb pressure wave.
3- For loaded faces of element, the pressure wave arrival time is calculated either by
Friedlander equation11) or by the discrete data. Friedlander equation11) gives;
t
P(t ) = Ps1 , P(t ) > 0
Ts
(5-1)
Where t is the time measured since wave arrival; P(t) the pressure value; Ps the peak
static overpressure at the wave front; Ts the duration of positive phase. These values can
be calculated from given charts and empirical equations shown in Ref. 3.
4- Elements are loaded only when the pressure wave front reaches the element faces.
Hence, elements near to the blast source are loaded first.
5- The pressure wave is assumed perpendicular to the loaded surface. The applied
force value is the multiplication of pressure value by element area and its
direction is perpendicular to the loaded face.
6- The structure is solved for the blast forces and then another time step is applied.
The time step applied for calculations is 0.001 seconds to follow the wave front
accurately.
The pressure value of the blast wave is normally very high close to the source and
reduces drastically when moving far from the source. On the other hand, the duration of
the pressure wave is very short close to the blast source and it increases when moving
away from the source. It should be noted that using free-field pressure wave models
indicates that the reflection and refraction of pressure wave at the ground surface and
surrounding buildings were not taken into consideration. More details about the pressure
wave models are given in Ref. 11.
In ELS you can set the time of explosion, bomb weight and position. Simply select the
type of bomb from the drop-down list box and enter the relevant information in the table
shown in Fig. 5.6
47
Fig. 5.6 Parameters defining bomb loads
5.2.7 Element Removal

This method is generally used in building implosion and progressive collapse cases,
where the user knows which elements are expected to fail and thus cause the structure to
collapse. In this case, the user may select the exact element(s) (building component) that
will be considered failed and thus immaculately removed. The advantage of using such a
way is the reduction in computation time compared to the blast solution. Select the
element or group of elements using the Select Elements" button. Click and drag a
window around the desired element or select a group of elements by holding the shift
button while drawing a window. Selected elements must be meshed prior to making the
selection.
48
Fig. 5.7 Parameters defining element removal
Fig. 5.8 Parameters defining pressure control
5.2.8 Pressure Control

Pressure history can be easily introduced to the structure in ELS as shown in Fig. 5.8.
The following must be defined:
1- Pressure history
49
2- Location of the source of the input pressure

As for the pressure history, the user needs to input the pressure values with discrete time
intervals. This could be also imported from an external text file. The location of the
source of the input pressure decides which elements are subjected to compression and
which are subjected to suction.
It is worthy to note that when the input pressure has an abrupt change in the value such as
in the case of the peak pressure of a bomb as shown in Fig. 5.9. Using larger time step
with large number of increments would lead to inaccurate pressure simulation as shown
in Fig. 5.9.b. A better solution is to use small time step with small number of increments
as shown in Fig. 5.9.c. For example it would be better to use T= 0.0001 seconds with
one increment than using T=0.01 seconds with 100 increments.
Pressure
Pressure
Pressure
Peak
Peak
Peak
Large and can not

be neglected
Small and neglected
a) Pressure history
Time
Time
Time
b) large DT with large number of

increments
c) Small DT with small number of

elements
Fig. 5.9 Pressure control with abrupt change in pressure value
50
6 Boundary and Initial Conditions

6.1 Analysis Domain Boundaries
To solve a problem numerically, there must be a displacement (geometry) domain, which
is the space that encompasses the body, and all the possible locations to which every
point on the body can move. Fig. 6.1 shows the analysis geometry domain and the
boundary of the modeled body. The points, along the boundary, where the body rests are
named supports. These supports have special motion conditions. The supports may be
restrained to move to any direction, or can be restrained to move at a specific direction
only. As these points are located on the boundary of the body, with special motion
conditions, they are named as boundary conditions.
Suppose a time dependant force, the analysis will be slightly complicated. Usually, the
duration of force application is divided into smaller steps called solution intervals. Each
interval starts with a force value and ends with another force value, which is considered
as the start force for the next solution interval. It can be easily concluded that not only the
geometry domain must be defined but also the time domain must also be defined for the
solution. For time dependant problems, the unknowns will be the displacements of the
body which are varied with time. The displacement rate is the velocity and the velocity
rate is the acceleration. The body displacement values (or time derivatives) at the initial
time (t=0.0) are called the initial conditions.
6.2 Assigning Boundaries in ELS

Assigning boundary conditions logically comes after meshing (dividing the body into
small elements) and after completing the process of load application at some elements of
the body. As mentioned earlier, analysis of the problem under consideration will be
solved in a domain. The analysis domain can be defined as the space that encompasses all
the possible positions of the body, during motion (dynamic analysis) or deformation
(static analysis). The analysis domain also encompasses the motion of all the fragments
and debris from crushed elements during the collapse process. Referring to Fig. 6.1, the
plan of minimum Z represents the ground boundary. For static analysis, every element
attached to the ground boundary is considered as fixed (i.e. no motion). On the other hand
by assigning a minimum limit in the Z direction for the dynamic analysis of a falling
body (dynamic motion towards negative Z-direction), the body will not move past this
limit. After reaching the minimum Z plan, the body may either rebound, crash into
fragments or stick to the boundary depending on the both the stiffness of the model and
the ground. Fig. 6.2, shows the inputs for assigning the analysis domain in ELS Software.
The ground boundary material can be selected as shown in Fig. 6.2. It can be also
expanded (Ground XY expansion ratio) as wide as needed to capture all the falling
objects especially in large scale building models collapses, where numerous fragments
and debris exist. In conclusion, the ground boundary is a support boundary.
51
Fig. 6.1 Definitions: Analysis Domain Boundaries, Boundary, Support
Fig. 6.2 Analysis Domain Boundaries in ELS Software
6.3 Assigning Boundary Conditions (Restrains or Supports)

Boundary conditions are assigned to a point of the body (in other words to the center of
mass of a small element in the body mesh). Selecting the element (or group of elements)
and referring to Fig. 6.3, the boundary conditions are either displacement restraints or
rotational restraints. In total, six restraints are available. The boundary conditions may
differ according to the loading stage, or may be the same for all stages. The boundary
conditions can be modified directly in the table for a specified element or rather entirely
removed.
6.4 Initial conditions (Velocity or Acceleration)

To solve a time-dependant problem numerically, initial conditions are required (velocity
and acceleration values at the start of motion, t =0.0). As a default, the body initial
52
conditions are set equal to zero as long as the motion starts from rest. However, in some
modeling cases such as modeling projectiles or elements in motion, the initial velocities
or accelerations may not be equal to zero.
Referring to Fig. 6.4, the initial velocity for some selected elements are assigned. Six
velocity components (three translational and three angular components) exist. The stage
of loading can also be selected. Stage 1 is the own weight loading stage as previously
explained. In this stage, the modeled body is considered to be at rest (fixed) as no motion
is yet started. Other loadings as static or dynamic loadings can be set in the following
stages. Fig. 6.3 shows the boundary conditions applied at stage 1. As shown in Fig. 6.4,
the initial velocity is assigned in (Stage 2). Same procedures can be followed when
assigning initial accelerations.
Fig. 6.3 Boundary Conditions:

[Elements No. 1,2,3,4 are assigned translational restraints in X, Z directions together with rotational
restraints around Y and Z directions. All these restraints are assigned for Stage 1]
53
Fig. 6.4 Initial Conditions: [Elements No. 15,17,18,19, 20 are assigned initial translational velocity in
negative X direction for loading Stage 2]
54
7 Elements contact
7.1 Introduction
One of the main break-through features in ELS, is automatic element contact detection.
The user is not have required to predict where or when contact will occur. Elements may
contact and separate, re-contact again or contact other elements without any kind of user
intervention. A lot of efforts were made to make this possible. This section shows an
overview of how the contact between elements are modeled.
7.2 Contact Types

There are several types of contacts; corner-to-face contact, edge-to-edge contact and
corner-to-ground contact.
7.2.1 Corner-Face type

Figure 7.1 illustrates the first types of contacts, in which a corner from one element hits
one face of another element. In some cases, as shown in Fig. 7.2, more than a corner of
each element may contact the other elements.
Shear spring in y
Shear spring in X
Normal Spring
Fig. 7.1 Corner- to-Corner Contact
Fig. 7.2 Element colliding with two elements
55
7.2.2 Edge-Edge Type
Con
spr tact s
hea
in g
r
As shown in Fig. 7.3, two elements are in contact at their edges.
r
ea
sh
t
ac
nt g
Co r i n
sp
Contact
normal spring
Fig. 7.3 Edge-toedge contact
7.2.3 Corner-Ground Type

Referring to Fig. 7.4, the corner of an element is in contact with the ground. This is very
similar to the type corner-face except that the contact is between the element and the
ground at Min-Z instead of having the contact with other element face.
Ground
Shear spring in y
Ground
Shear spring in X
Ground
Normal Spring
Fig. 7.4 Corner- to-Ground Contact
7.3 Contact Stiffness

The following steps are used to include the contact stiffness matrix:
a) Getting the Contact point. In case of corner-face type, the corner coordinate is
considered as the contact point, while in edge-edge type, the center of the line
connecting the closest two points of each edge is considered the contact point
b) Adding contact springs. Three contact springs are added at each contact point;
one normal spring and two shear springs. In case of corner-face type or cornerground type, the normal spring is considered perpendicular to the contact face
while in case of edge-edge type, the normal spring is perpendicular to the plan
composed of the two contact edges. The two shear springs are perpendicular to
the normal spring and to each other. In case of corner-face type, the shear springs
are in the same plan with the element face while in case of edge-edge type; the
shear springs lie in the same plan with the two colliding edges
56
c) To calculate the contact stiffness this rule is applied. When two elements with
different material properties collide, the spring properties are governed by
material with softer properties. The following explains how contact stiffness is
included:
i. Referring to Fig. 7.5 & 7.6, assume that the contact area between
the two elements is a ratio of the average element dimension. This
ratio is NF for normal stiffness and SF for shear stiffness
ii. The normal stiffness = E * Contact Normal Area/D = E * NF
*D*D/D= E*NF*D where E is the minimum Youngs Modulus of
the two bodies, NF is the Contact Normal Stiffness ratio and D is
the average element dimensions.
iii. The shear stiffness = E * Contact Shear Area/D = E * SF
*D*D/D= E*SF*D where E is the minimum Youngs Modulus of
the two bodies, SF is the Contact Normal Stiffness ratio and D is
the average element dimension for the element.
7.4 Energy Dissipation During Contact

During collision of elements, some kinetic energy of elements before collision is lost
during collision process. The rebound factor (r) is defined as the ratio between the
element relative velocity before and after collision. The value of "r" ranges between zero
and one. Having "r" equal to one means that the relative velocity after collision is the
same as before collision and hence, no energy dissipation. While, all kinetic energy is
lost during contact if "r" equals zero.
In general, representation of energy dissipation during the collision process is performed
by determining the velocity of the elements after collision from the momentum theory
and using rebound factor "r". Although this technique is simple, when it is applied to
continuum material like concrete, a small time increment is required to simulate
transmission of stress waves due to the collision through other elements.
Contact area (A)
D
NF*k
Normal spring stiffness = E * A/ D

A = NF D2
Default value of NF= 0.01
Fig. 7.5 Normal contact spring factor
57
Contact
area (A)
SF*k/1000
D
SF*k
Shear spring stiffness = G * A/ D

A = SF D2
Default value of SF= 0.001
If shear force <

friction coefficient
x Normal force,
If shear force >

friction coefficient
x Normal force,
shear stiffness
= G * SF * D.
shear stiffness
is G * SF * D /
1000.0
Fig. 7.6 Shear contact spring factor
The following technique is proposed for representation of the energy dissipation during
contact. The load-displacement relation of a spring during loading (approaching) and
unloading (leaving) are shown in Fig. 7.7. Assume the factor "n" represents the ratio
between the unloading and loading stiffness. The value of "n" should be greater than "1".
Having "n" equals 1 means that there is no energy dissipation during contact process
while all the kinetic energy is lost if "n" value approaches infinity. The unloading
stiffness factor "n" can be correlated easily to the rebound factor "r" by equating the
rebound energy of elements in both techniques. This relation is:
r=
1
n
(7-1)
This indicates that energy dissipation during contact can be simulated by any of these
methods. The main advantage in using the unloading stiffness technique is that the time
increment used needs to be reduced only during the contact of elements. A larger value
of time increment can be used after separation.
Fig. 7.7 Load-displacement relation of a contact spring in loading and unloading condition.
58
7.5 Time step

It should be noted that the time step required for stable dynamic analysis is a function of
element mass and contact stiffness. When the contact is soft or element mass is high,
longer time steps can be used for simulations while hard contacts or having low element
mass results in the necessity to use smaller time steps.
In case of analysis of reinforced concrete, elements are usually connected by steel bars
and stirrups and hence, when contact occurs with the ground, a bulk of elements are
colliding at once and hence, larger time steps, like 0.001 seconds, can be used for the
analysis which leads to less computational time. However, in case of very brittle material,
like plain concrete and glass, elements tend to completely separate and collide with the
ground while not attached to other elements and hence, smaller time steps, like 0.0001
seconds, should be used for these types of simulations.
59
8 Common Modeling tricks

8.1 Contact Problem Guidance
It is preferable to use small time increment when it is expected that contact between
elements will take place.
Example:
Figure 8.1-a shows a block moving with a velocity of 60 m/sec hitting a vertical wall.
When using a time increment of 0.01 it is noticed that the contact between the block and
the wall could not be detected. While in Figure (1.b) the block impact was detected using
a time increment of 0.001.
Fig. 8.1-a block impacting a vertical wall (Large time increment)
Fig. 8.1-b block impacting a vertical wall (short time increment)
*Reference samples (wall-pass.ELS & wall.ELS)
8.2 Common stability problems

8.2.1 Static Loading
It is preferable to fix the boundary conditions of the element bottom if this bottom is
higher than the ground boundary plan (Min. Z).
Example: Referring to Fig. 8.2, in the left column, the column base (bottom end) is lying
on the ground boundary while in the right column; the column base is higher than the
60
ground boundary. If the column base is located on the ground boundary, it will be treated
as a fixed end. On the contrary, if its base is higher than the ground boundary, boundary
conditions should be assigned to the base otherwise this end would be treated as a free
end.
8.2.2 Dynamic Loading

In some dynamic loading cases, as for falling objects, it is preferable to assign the
boundary condition of the first stage to be fixed while no boundary conditions are
assigned in the other (i.e. free from any restrains). Figure 8.3 shows that all the elements
of this block are fixed in stage 1.
Fig. 8.2 Column base boundary
Fig. 8.3 Free falling block
8.3 Guidance for modeling Reinforcing Bars

8.3.1 Reinforcing Bars Extension Length
If the extension length of the bars to the adjacent elements is not appropriate, the bar
connectivity might not work properly. Figure 8.4 shows the analysis results of two
beams; each consisted of three connected sub-beams. The middle sub-beam of the first
one has a bar extension length equals zero, while that of the other beam has a positive
extension length. The first one as a result, gave wrong results with wide cracks, near
element separation exist at the sub-beams connection zone. On the other hand, when
using an appropriate bar extension length, the sub-elements did not separate from each
other.
61
(a) Extension length is not applied
(b) Extension length is applied
50
45
Bars extended
40
Bars not extended
Load (Ton)
35
30
25
20
15
10
5
0
0
Displacement (cm)
(c) Comparison of behavior of beams with and without bar extension

Fig. 8.4 Reinforcing bars extension effect
Reference: samples (beam.ELS, beam extension.ELS)
8.3.2 Modeling Of Non Symmetric Bars In Columns

When modeling columns with non-symmetric reinforcement, it is advised to create a
virtual RFT column inside the concrete column and to make its behavior as RFT, the steel
bars in the virtual column, that are located inside the concrete column, will be added to
the steel bars in the symmetric concrete column as shown in Fig. 8.5. The steel bars on
the other end of the virtual column will not be considered unless intersecting elements.
62
Fig. 8.5 Non-symmetric reinforcing bars in columns
8.4 Properly Including Own Weight

The first stage of loading in most cases of structural simulation is the loading of the
models own weight. This must be performed before the application of any other loads to
the model so as not to affect the component properties unrealistically. This is
automatically executed by means of the solver and Stage 1.
Reference: samples (stadium.ELS, half_ distraction.ELS, 2building.ElS)
8.5 Post-Peak Behavior Achievement

If grasping the post peak behavior is of the user interest, it is advised to use the
displacement control option rather than the load control option. Figure 8.6 shows the
difference in results when using load control and displacement control.
60
50
Peak Point
Load (Ton)
40
30
20
10
Load Control
Displacement Control
0
0
Displacement (cm)
Fig. 8.6 Post-peak behavior
Reference: samples (ldcontrol.ELS, discontrol.ELS)
63
8.6 Contact Springs Unloading Stiffness Factor

Among the material properties used there is a factor termed Unloading Factor. This
factor describes the dissipation of energy due to impact. The higher this factor is, the less
the rebound. Figure (8.7.a) shows the behavior when using unloading factor equal to ten
while Fig. (8.7.b) Shows the behavior when using unloading factor equal to one.
Fig. 8.7-a Unloading factor=10
Fig. 8.7-b Unloading factor=1
Reference: samples (steel fall 10.ELS, steel fall.ELS)
8.7 Cyclic Point loading

For correct analysis of cyclic loading, it is preferable to use an elastic plate at the loading
point in order to prevent cracking localization in concrete at loading point. Figure 8.8
illustrates that effect, in which it could be seen that a high localization takes place for the
point loading on the concrete. This actually happens when the load is in the direction that
causes tensile stresses in the loading point concrete.
64
25
25
20
20
15
15
10
10
5
0
-0.3
-0.2
-0.1
-5 0
0.1
0.2
0.3
Load (Ton)
Load (Ton)
5
-3
-2
-1
0
-5 0
-10
-10
-15
-15
-20
-20
-25
-25
Displacement (cm)
Displacement (cm)
(a)Behavior when applying the cycles on one

element only at the side
(b) Behavior when applying the cycles on

one element only in the middle of the plate
Fig. 8.8 Cyclic point loading
Reference: samples (cyclic col. ELS , cyclic col right.ELS)
65
9 Verification Samples
The verification samples for different types of loading are as follow;
9.1 Static Loading

9.1.1 Elastic stage
Reference: samples/static/ elastic
Table 9.1 Verification samples of static loading in the elastic stage.
File name
Description
Simple_Beam.ELS
This sample describes the analysis of a simple beam

under static uniform load in the elastic stage.
Cantilever_frame.ELS
This sample describes the analysis of a cantilever frame

under static Point load in the elastic stage.
Cant_slab.ELS
This sample describes the analysis of a cantilever slab

under static uniform load in the elastic stage.
Simple_slab.ELS
This sample describes the analysis of a simply supported

slab under static uniform load in the elastic stage.
9.1.2 Non-linear Stage

9.1.2.1 Monotonic loading
Reference: samples/static/ nonlinear/monotonic.
66
Table 9.2 verification samples of static loading under monotonic load.
File name
Description
culvert.ELS
This sample describes the analysis of a reinforced

concrete culvert under static monotonic load.
Deep_beam1.ELS

concrete Deep beam without web reinforcement subjected
to 4 points loading.
Deep_beam2.ELS

concrete Deep beam with web reinforcement subjected to
4 points loading.
Slab.ELS

concrete slab under static monotonic point load.
Squat_uni_shear.ELS
This sample describes the analysis of a squat column

under combined axial load and shear load.
ELSS0.ELS
This sample describes a nonlinear analysis of steel plate

under monotonic loading
ELSS3.ELS
This sample describes a nonlinear analysis of a hot-rolled

steel beam under flexure
Steel_frame.ELS
This sample describes a nonlinear analysis a nonlinear

analysis of hot-rolled steel frame under monotonic
loading
Comp_slab.ELS
&Comp_slab_b3.ELS
This sample describes a nonlinear analysis Steel

Composite Beam under Four-Point Loading
hybrid steel girder.ELS
This sample describes a nonlinear monotonic analysis of

hybrid steel girder
comp Girder.ELS
This sample describes a nonlinear analysis of a concretefilled tube girder under four-point loading
armoti.ELS
This sample describes a nonlinear analysis of stoneconcrete bearing walls subjected to lateral loads
CMU_FRP.ELS
This sample describes a nonlinear analysis un-reinforced

masonry wall under out-of-plane loads
window_glass.ELS
This sample describes a nonlinear analysis of a glass

window subjected to lateral loading
9.1.2.2 Cyclic loading
Reference: samples/static/ nonlinear/cyclic
67

Table 9.3 Verification samples of static loading under cyclic load.
File name
Description
slab_cyclic.ELS
This sample describes the analysis of a slab under cyclic

load.
Col_cyclic.ELS
This samples describes the analysis of a column under

reversed cyclic loading. The column fails is shear
ELSS2.ELS
This sample describes a nonlinear cyclic analysis of Steel

Plate
9.2 Dynamic loading

9.2.1 Elastic Stage
Reference: samples/Dynamic/ elastic
Table 9.4 Verification samples of dynamic loading in the elastic stage.
File name
Description
2DOF.ELS
This sample describes the free vibration of two story shear

building (two degrees of freedom)
projectile.ELS
This sample describes the motion of a projectile under

gravitational acceleration.
SDOF.ELS
This sample describes the free vibration of a column with a

top mass in the elastic stage (single degree of freedom)
9.2.2 Non-linear Stage

Reference: samples/Dynamic/ nonlinear
Table 9.5 Verification samples of nonlinear dynamic loading.
File name
Description
pier.ELs
This sample describes non-linear seismic response of RC

pier subjected to eccentric axial force
9.2.3
68
9.2.4 Demolition and Blast

Reference: samples/Demolitions
Table 9.6 Verification samples of demolitions.
File name
Description
Chimny.ELS
Demolition process of a masonry chimney
Partial_Demolition.ELS
Demolition of reinforced concrete multi story building
STADIUM.ELS
Demolition of reinforced concrete stadium
steel tank.ELS
Demolition process of a steel tank
69
10 Case Studies
The following case studies are also included in the ELS;
Table 10.1 Case studies using ELS
70
File name
Description
RC4-Stories.ELS
Two column removal cases simulating the resulting

behavior of the structure showing progressive collapse
side sway.ELS
Steel frame in a 2D environment with loads applied at the

top opposite ends resulting in buckling of the steel
columns
steelRemoval.ELS
Three bottom columns are removed from the front of the

steel frame resulting in the collapse of the structure
steelStress.ELS
Lateral loads are applied to the side of the steel structure

resulting in the frames deflection
Window Blast.ELS
Explosive placed in front of one-story RC structure with

masonry walls and glass windows ending in severe
damage to the walls and windows
11 References
1. Kimiro Meguro and Hatem Tagel-Din: Applied Element Simulation of RC
Structures under Cyclic Loading, ASCE, Vol. 127, Issue 11, pp. 1295-1305,
November 2001.
2. Kimiro Meguro and Hatem Tagel-Din: Applied Element Method for Structural
Analysis: Theory and Application for Linear Materials, Structural
Eng./Earthquake Eng., International Journal of the Japan Society of Civil
Engineers (JSCE), Vol. 17, No. 1, 21s-35s, April 2000.
3. Hatem Tagel-Din and Kimiro Meguro: Applied Element Method for Simulation
of Nonlinear Materials: Theory and Application for RC Structures, Structural
Eng./Earthquake Eng., International Journal of the Japan Society of Civil
Engineers (JSCE) Vol. 17, No. 2, 137s-148s, July 2000.
4. Hatem Tagel-Din and Kimiro Meguro: Applied Element Method for Dynamic
Large Deformation Analysis of Structures, Structural Eng./Earthquake Eng.,
International Journal of the Japan Society of Civil Engineers (JSCE), Vol. 17,
No. 2, pp. 215s-224s, October 2000.
5. Kimiro Meguro and Hatem Tagel-Din: AEM Used for Large Displacement
Structure Analysis, Journal of Natural Disaster Science, Vol. 24, No. 2, pp. 6582, 2002.
6. Hatem Tagel-Din: Collision of Structures During Earthquakes, Proceedings of the
12th European Conference on Earthquake Engineering, London, UK, September
9th - September 13th, 2002.
7. Hatem Tagel-Din and Kimiro Meguro: Analysis of a Small Scale RC Building
Subjected to Shaking Table Tests using Applied Element Method, Proceedings of
the 12th World Conference on Earthquake Engineering, New Zealand, January
30th -February 4th, 2000.
8. Kimiro Meguro and Hatem Tagel-Din: A New Simplified and Efficient
Technique for Fracture Behavior Analysis of Concrete Structures, Proceedings of
the Third International Conference on Fracture Mechanics of Concrete and
Concrete Structures (FRAMCOS-3), Gifu, Japan, Oct. 1998.
9.
Okamura H. and Maekawa K.: Nonlinear analysis constitutive models of

reinforced concrete, Gihodo Co. Ltd., Tokyo, 1991.
10. Ristic, D., Yamada, Y., and Iemura, H. (1986), Stress-strain based modeling of
hysteretic structures under earthquake induced bending and varying axial loads,
Research report No. 86-ST-01, School of Civil Engineering, Kyoto University,
Kyoto, Japan
11. Mays G.C. and Smith P.D.: Blast Effects on Buildings, Thomas Telford
Publications, Thomas Telford Services Ltd, Heron Quay, London, UK, 1995.
12. Compaq Extended Math Library Reference Guide, Compaq Computer
Corporation, Houston, Texas, 2001.
71

Technical Manual

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Technical Manual

Uploaded by

Copyright:

Available Formats

Extreme Loading for Structures

Extreme Loading is made possible by

Extreme Loading for Structures Technical Manual

Applied Science International

2006 Applied Science International, LLC

Extreme Loading for Structures Technical Manual

Extreme Loading for Structures Technical Manual

Extreme Loading for Structures Technical Manual

APPLIED ELEMENT OVERVIEW................................................................................................. 5

MATERIAL MODELS AND FAILURE CRITERIA ................................................................... 32

ELASTIC MODEL ........................................................................................................................ 32

WHAT ARE LOADING STAGES?.................................................................................................... 39

Extreme Loading for Structures Technical Manual

BOUNDARY AND INITIAL CONDITIONS................................................................................. 51

COMMON MODELING TRICKS ................................................................................................. 60

ANALYSIS DOMAIN BOUNDARIES .............................................................................................. 51

ELEMENTS CONTACT ................................................................................................................. 55

Load Control ........................................................................................................................ 44

CONTACT PROBLEM GUIDANCE ................................................................................................. 60

Extreme Loading for Structures Technical Manual

Extreme Loading for Structures Technical Manual

- How does falling debris affect adjacent structures?

Fig. 1.1 Analysis domain of AEM

Extreme Loading for Structures Technical Manual

Table 1.1 Application fields of AEM

Degrees of freedom 6 per element

16 per element (8-node element)

Smeared cracks and physical

Smeared cracks and physical

No need for joint elements

Joint elements (interface

No need for joint elements

Joint elements (interface

Cracks propagates easily at

Cracks can not extend from

Extreme Loading for Structures Technical Manual

Crack propagation can be

Crack propagation cannot be

All reinforcement details, for

Very difficult to accurately

During and after

Can not follow collapse behavior

Extreme Loading for Structures Technical Manual

2 Applied Element Overview

a. Element generation for

b. Spring distribution and area of influence of each

Fig. 2.1 Modeling of structure to AEM

Extreme Loading for Structures Technical Manual

Fig. 2.2 Nodal compatibility in FEM

Extreme Loading for Structures Technical Manual

Extreme Loading for Structures Technical Manual

b- Simulated cracks in FEM

2.2.1.2 Discrete Crack Approach

Extreme Loading for Structures Technical Manual

Predefined Discrete Cracks

2.2.1.3 Moving Mesh Techniques

2.2.2 Element connectivity at AEM

Extreme Loading for Structures Technical Manual

Allowed (Some springs may fail

Separation of nodes causes stress Separation of springs does not

Transition elements are needed

No need for transition elements

Very complicated specially at

Every object is meshed alone