You are on page 1of 86

Orion Training

Presentation

Copyright CSC(UK) Ltd. 2006


Orion Training
Orion Training

Shear Wall Modelling

Copyright CSC(UK) Ltd. 2006


Shear Wall Modelling

CONTENT
Introduction
Simple 2D Wall Panel
2D Wall Panel With Openings
3D Core Wall
Interaction of Beams & Walls
Examples in Orion
Overview and Conclusions
Introduction
This presentation expands upon a Technical Note Published in
“The Structural Engineer” 1st February 2005, Vol 83 No.3, page
20.
It can be downloaded from the CSC website – www.cscworld.com
Several other articles, papers etc prepared by CSC staff are
available from the web site.
Introduction
The Technical Note was titled:
NEW MODELLING, SAME ANSWERS
In writing this the objectives were:
To emphasise that FE modelling does not give new
answers.
The same answers can be achieved using FE or traditional
Beam Element Idealisations.
To raise awareness of issues introduced as you
begin to model with FE (Shell) elements.
That in fact more care is needed to get acceptable
answers using FE.
That there is probably more chance of making serious
mistakes using FE.
Introduction
Do we actually know what the “correct”
answers are?
A quote from another Technical Note:
“Engineering (and some may think FE practice
also) is the art of modelling materials we do not
wholly understand, into shapes we cannot precisely
analyse, so as to withstand forces we cannot
properly assess, this in such a way the public and
(hopefully) the customer has no reason to suspect
the extent of our ignorance.”
Structural Engineering Modelling and Analysis,
by Arthur T. Murphy
The Structural Engineer – 3rd Feb 2004
Introduction
What does this mean in the context of
concrete shear walls?
If deflections and the distribution of forces are
important to you then you need to get all
aspects of your model right, this means:
Accurate material properties for each member.
Short term E for C40 somewhere between 22 and
34kN/mm2, then what about creep allowance?
Accurate section properties for each member.
Allowance for flanges and cracking?
A good arrangement of members to idealise the
overall physical geometry.
This is what we are concentrating on in this presentation.
Introduction

Remember:
We are engineers.
Engineers routinely idealise structures in
order to make reasonable (safe) estimates
of design forces.
We should not expect or demand exact
correlation between results achieved by
using different structural idealisations.
Orion Training
Simple 2D Wall Panel
Can Wall Panels be
idealised with Beam
Elements?
To be convinced that
they can we need to
start with something
simple - consider the
wall shown here.
Simple 2D Wall Panel
What Deflection would you
expect?
Deflection = F L3 / 3 E I
(Ignoring shear deformation)
F = 100kN
L = 35m
E = 13E6 kN/m2
I = 3.6m4
Deflection
= 100 x 353/(3 x 13E6 x 3.6)
= 30.5mm
So allowing for shear
deformation should expect
something a little greater than
30.5mm
Simple 2D Wall Panel
Simple 2D Wall Panel
Beam 1 Shell 9 Shell 36 Shell 81 Shell

Lateral Deflection 31.2 29.7 31.0 31.2 31.2


(mm)

5 different models of the same wall.


As meshing increases result converges on that given
by the simplest beam element idealisation.
First hint at importance of meshing and mesh
sensitivity.
And the beam idealisation has other advantages….
Simple 2D Wall Panel
In order to design a wall panel you need to know
“Panel Forces”.
Simple 2D Wall Panel
For a beam element idealisation Panel Forces
drop out naturally:
Panel Forces = Member Forces
For a meshed wall the panel force is
determined by re-integration of shell results
along a cut line.
S-frame has tools allowing you to do this
interactively.
Orion will automatically re-integrate results to
generate design forces at each floor level.
Simple 2D Wall Panel
A similar model in Orion…..
Simple 2D Wall Panel
Simple 2D Wall Panel
Simple 2D Wall Panel
Simple 2D Wall Panel
Conclusions on Simple 2D Wall Panel
The beam element idealisation is excellent.
Results are potentially better than those that
would be achieved by using a poorly meshed FE
model.
Results are more readily usable than those given
by an FE model.
Orion Training
2D Wall Panel With
Openings
Some engineers may be surprised at the
accuracy of the good old fashioned
beam model in the previous example but
still feel that the only way to take
account of openings in walls is to resort
to the use of shells.
In this example we will look at the same
wall panel, with the same loads applied,
but with significant (door) openings cut
out of the wall at every floor level.
2D Wall Panel With
Openings

First we will examine the results obtained from


a series of increasingly finely meshed shell
models as shown above.
2D Wall Panel With
Openings
Model Sway (mm) Section A-A Section B-B

Axial (kN) Shear (kN) Moment Axial (kN) Shear (kN) Moment
(kNm) (kNm)

7 Shell 33.6 269 35 194 17 69 66

63 Shell 36.0 256 40 225 12 60 58

252 Shell 36.6 253 41 232 11 58 57

567 Shell 36.8 252 41 235 10 58 56

Notice convergence as meshing is refined


Also note reasonably significant variation in
section design forces.
2D Wall Panel With
Openings
In the Technical Note we looked at two styles of model as shown
below.
We noted the importance of attaching the coupling beam to the
mesh but did not demonstrate this issue in detail. Let’s consider
that in a little more detail now…
2D Wall Panel With
Openings
Coupling Beam modelled
with beam (not meshed)
Analysis Without “DDF”
Deflections Huge (414mm) –
nothing like the expected
37mm.
No moments develop in
coupling beams.
Obviously wrong – there is
no moment interaction
between the beams and
the meshed wall panels.
2D Wall Panel With
Openings
Analysis With “DDF”
Deflections much smaller
As the mesh is refined the peak
deflections do not converge.
37.7mm
46.8mm
60.9mm
118.0mm
The moments in the coupling
beams also vary – lower moments
where mesh is finer.
Obviously wrong ?
You might only notice if you
undertake a mesh sensitivity
assessment.
2D Wall Panel With
Openings
Connection Detail
Left – the coupling beam induces
large local stresses (and distortions)
in the fine mesh.
Right – by modelling the physical
extent of the connection this stress
concentration is reduced.
2D Wall Panel With
Openings
Some initial surprise that the DDF potentially produces such poor results.
There are other ways to model the interconnection of the beam and the
meshed wall, we will introduce that later.
For now we will continue with the models and comparisons made in the
technical note…
2D Wall Panel With
Openings
Mid-Pier Model
Wall is not meshed, it is idealised
by a single beam element
positioned centrally in the panel
(hence “mid-pier”)
Coupling beams are attached to
the face of the wall and designed
for moments that occur at this
location.
So the connection between the
coupling beam and the mid-pier
(wall) beam has to be made with
something else - a rigid beam.
Suggested properties of rigid
beam elements are given in the
technical note – in this example
they are taken as 10 to 100
times stiffer than a 3.5m deep
beam.
2D Wall Panel With
Openings
4 models were considered:
Models 1 and 3 have meshed wall panels, models 2 and 4 use the mid-pier idealisation.
Models 1 and 2 have the coupling beams modelled at mid-height.
Models 3 and 4 are essentially the same, but the coupling beams are moved up and
modelled at the top level of the beam.
Results can now be compared.
2D Wall Panel With
Openings
Model Sway (mm) Section A-A Section B-B

Axial (kN) Shear (kN) Moment Axial (kN) Shear (kN) Moment
(kNm) (kNm)

7 Shell 33.6 269 35 194 17 69 66

63 Shell 36.0 256 40 225 12 60 58

252 Shell 36.6 253 41 232 11 58 57

567 Shell 36.8 252 41 235 10 58 56

Model Sway (mm) Section A-A Section B-B

Axial (kN) Shear (kN) Moment Axial (kN) Shear (kN) Moment
(kNm) (kNm)

Model 1 36.4 256 37 229 14 57 55

Model 2 36.2 256 50 239 0 61 61

Model 3 36.5 255 40 234 11 57 53

Model 4 36.2 251 50 249 0 67 67


2D Wall Panel With
Openings
Discussion of Results:
MODEL 1 – Meshed wall panels to each side of the opening with
a beam element used for the coupling beam positioned on its
physical centreline.
The results for this model agree well with the results of the fully
meshed versions of the model.
MODEL 2 – Mid-Pier wall model with a beam element used for
the coupling beam positioned on its physical centreline
The results in table 3 indicate that this simple model also compares
well to the meshed models in table 2.
The only exception is that the shell models indicate an axial load
within the coupling beam while the beam models do not. This is
explicable and relates entirely to the vertical load in each meshed
panel – when a meshed panel is compressed vertically the sides of
the panel expand laterally (as dictated by Poisson’s ratio) and these
opposing lateral expansions are resisted by the coupling beam. This
effect is negligible and would normally be ignored therefore this
difference is not considered significant.
2D Wall Panel With
Openings
Discussion of Results (cont’d):

MODELS 3 & 4 – Repeats of models 1 and 2 but with the


coupling beams and rigid arms lifted and idealised at the top of
the coupling beam rather than on its centreline.
Clearly this is a less accurate idealisation but it is often much more
convenient to model everything in one floor at a common level and
top of structure is often chosen for this purpose.
It is interesting to confirm that this common idealisation has very
little impact on the results.
2D Wall Panel With
Openings
A similar model in Orion…..
Simple 2D Wall Panel
2D Wall Panel With
Openings
2D Wall Panel With
Openings
2D Wall Panel With
Openings
Modelling Note
Previously showed
modelling at beam – wall
interface with rigid links
extending up/down the
wall.
In Orion we have taken the
links through the wall.
There are various reasons
for this. The properties of
these links have been tuned
by testing of numerous
cases – more on this later.
2D Wall Panel With
Openings
Conclusions on 2D Wall Panel with Openings
Once again the beam element idealisation is very
good.
The results where meshed panels interact with
beam elements can be good provided attention is
given to the way the connections of beam and shell
elements are modelled.
IMPORTANT
Issues where beam elements interconnect with shell
elements are not restricted to this modelling problem,
they apply every time you try to mix beam and shell
elements.
Enhanced Shell elements that support the “DDF” should
be used cautiously, they do not necessarily provide a
solution.
Orion Training
3D Core Wall
Once again, some engineers may be
surprised at the accuracy of the mid-pier
(beam) model in the previous example
but still feel that the only way to deal
with core walls is to resort to the use of
shells.
In this example the mid-pier model will
be shown to extend very successfully to
the analysis of 3-dimensional core walls .
3D Core Wall
We will consider here a simple C-
shaped core slightly offset from
the centre of a simple 10 storey
building
The columns around the core are
held in position by floor diaphragm
action.
In this model two load cases are
considered.
Sway in X – loads are applied in
the X direction (as shown - parallel
to the flanges of the core).
Sway in Y – similar loads are
applied in the Y direction – since
the effective line of action of this
load is eccentric to the centre of
resistance provided by the core we
expect this case to result in
twisting at each floor level.
3D Core Wall
As before we can look at the model using a series of
increasingly finely meshed versions of the core and compare
these results with those given by a beam idealisation.
Model Loads applied in X Loads applied in Y

Max Sway End Panel Forces Max Sway End Panel Forces
in X (mm) in X (mm)
Axial Moment Shear Axial Moment Shear
kN kNm kN kN kNm kN

Meshed model – 1
1 shell over floor
127.2 -2883 3886 527 +/- 373.7 39 5040 890
to-floor height.

Meshed model – 3
3 shells over floor 130.7 -3030 3659 520 +/- 381.6 -49 4803 843
to-floor height.

Meshed model – 9
9 shells over floor 131.4 -3061 3609 508 +/- 383.5 -74 4743 810
to-floor height.

Beam Model
132.6 -2963 3763 525 +/- 381.9 33 4892 848
3D Core Wall
Discussion of Results:
Once again the beam model is in excellent agreement.
With loads applied in the X direction all the models sway without
twisting.
When an eccentric load is applied in the Y direction all the models
sway and twist. As expected the twist is symmetrical as is indicated
by the +/- values of sway in X.
In the mid-pier model tensions and compressions are developed
in the component panels of the core, this can only happen if the
rigid arms that link the mid-pier elements are effective.
3D Core Wall
A similar model in Orion…..
Simple 2D Wall Panel
3D Core Wall

Load in X direction
gives symmetrical
sway as expected.
3D Core Wall

Same Load Case


Meshed walls
give very similar
results.
3D Core Wall
Load in Y
direction
causing sway
and twist.
3D Core Wall
NEED CORRECT VIEW OF MESHED MODEL
Same Load Case
and again the
result is very
similar.
3D Core Wall
Wall Panel Forces
For design purposes
you ideally need to
know Panel Design
Forces.
Contouring looks good,
but does not yield this
information easily – re-
integration of results is
required.
The Forces in the
members of the mid-
pier model give what
you want with no
further processing.
3D Core Wall
Wall Panel Forces
Some general analysis
software such as S-Frame
provides tools to allow you
to define integration lines.
In Orion Integration Lines
are automatically inserted
at the top and bottom of
every wall panel.
The results shown here are
for the same wall panel as
noted earlier, meshed
model results at top, mid-
pier at bottom.
There is a 5 to 10 %
variation.
Orion Training
Interaction of Beams &
Walls
2D Wall Panel with Openings
We showed that the rigidity of the
connection between the beam and the
mesh needed to be modelled with care.
We noted that the use of shells supporting
the “DDF” is not in itself a robust solution.
3D Wall and Beam models
The above is not a 2D phenomenon, it can
be demonstrated in simple 3D models….
Interaction of Beams &
Walls
Consider a model
as simple as this.
We will apply a
horizontal and a
vertical loadcase.
Interaction of Beams &
Walls
Meshed Models (“Correct Answers?)
Lateral Load Case
Increased refinement – stable results
Deflection approx 5.0mm
Beam Moments can be determined by re-integration
End Moments = +/- 29kNm
Interaction of Beams &
Walls
Meshed Models (“Correct Answers?)
Vertical Loadcase
Increased refinement – stable result
Deflection approx 4.0mm
Beam Moments can be determined by re-integration
End Moments = 16kNm (hogging)
Mid Span moment = -59kNm (sagging)
Interaction of Beams &
Walls
Meshed Wall + Beam Model
No rigid links at connection.
Lateral Load Case
Increased refinement – deflections keep increasing.
Deflection approx 5.4mm for coarsest mesh and 7.5mm
for finest.
Diverging away from target of approx 5.0mm
Interaction of Beams &
Walls
Meshed Wall + Beam Model
No rigid links at connection.
Lateral Load Case
Beam Moments reduce as meshing is refined:
End Moments vary from 21kNm to 13.7kNm
Diverging away from target of approx 29kNm
Interaction of Beams &
Walls
Meshed Wall + Beam Model
No rigid links at connection.
Vertical Loadcase
Same issues exist.
Results diverging away from target deflections and
moments.
Interaction of Beams &
Walls
Meshed Wall + Beam Model
With rigid links at finer mesh connections.
Lateral Load Case
Deflections results are stable at 5.1mm.
Good agreement with 5.0mm given by fully meshed
model.
Interaction of Beams &
Walls
Meshed Wall + Beam Model
With rigid links at finer mesh connections.
Lateral Load Case
Beam Moments are also stable at 26kNm.
Reasonable agreement with 29kNm given by fully meshed
model.
Interaction of Beams &
Walls
Meshed Wall + Beam Model
With rigid links at finer mesh connections.
Vertical Loadcase
Beam Moments are again stable and in reasonable agreement
with values taken from the fully meshed model.
End Moments = 19.5 (compared to 16kNm)
Mid Span moment = -56 kNm (compared to -59kNm)
Interaction of Beams &
Walls
Automated Modelling in Orion
Problem
If we start adding vertical rigid arms
everywhere that a beam meets a wall we
will create a meshing nightmare!
We needed another solution.
Rather than extending down the wall we
have taken the rigid arms along the wall.
Interaction of Beams &
Walls
Orion Training
Examples in Orion
If time permits these examples will be
demonstrated live.
The following screenshots summarise
the points being made.
Examples in Orion
Symmetrical Model with unsymmetrical
modelling assumptions.
Examples in Orion
Examples in Orion
Model With Angled Line of Symmetry.
Loading applied along angled line.
Examples in Orion
Examples in Orion
Examples in Orion
Core Wall With Multiple Offset Openings.
Examples in Orion
Meshed Model
(course mesh?)
Loading applied in Y
direction causing sway
and twist – resultant
deflections shown in
mm.
A more finely meshedExamples in Orion
model shows that
previous deflections
were reasonable.
In this view the
moments developed in
the coupling beams are
shown.
Examples in Orion
Mid-Pier model results
for the same load case
cause very similar
sway and twist –
resultant deflections
are 4% greater.
The design moment Examples in Orion
view is dominated by
the moments
developing in the mid-
pier elements.
However, the values
indicated in the
coupling beams are
very similar to those
seen in the meshed
model.
For example, at first
floor 97.36 shown
here compares with
91.73 in the meshed
model – a 6%
variation.
Orion Training
Overview & Conclusions
Shear Walls and Core Wall Systems can be
modelled with relatively equal accuracy with
beam or shell idealisations.
Modelling with FE shell elements:
Is not inherently free from modelling issues /
difficulties.
Does not inherently produce better results.
Modelling with beam elements:
Probably requires more care.
And, the very fact that you have considered and
decided on idealisations tends to develop a feel for
the model and an expectation of the results. It is
more likely that unexpected results will be spotted
and investigated.
Overview & Conclusions

These sorts of conclusions are not


unique to shear and core walls.
Whenever you mix beam and shell elements
in any model there are potential issues.
Whenever you use shells at all you need to
assess sensitivity to mesh refinement.
In simplest terms, the old analysis adage
“RUBBISH IN = RUBBISH OUT”
is as likely to be applicable to shell
models as beam models.
Overview & Conclusions

Observation of General FE Software in


Use:
We have certainly seen a continuing increase in the
use of FE.
We have the feeling that once complex looking
models are developed using shells the results are
trusted a little too readily.
There are engineers using (or being asked to use)
FE with little previous training or understanding and
no time allocated to investigate and develop that
knowledge.
Overview & Conclusions
General analysis software allows you to build
your own models and take full responsibility
for your own modelling.
Building Modelling Products such as Orion are
beginning to automate the generation of the
underlying analysis models.
However, you still have a responsibility to
understand the modelling and check the results.
For Shear and Core Walls in Orion:
By supporting alternative methods we are providing a
logical way to help cross check results.
And importantly, the mid-pier modelling option is often
much quicker. On larger projects you might use it for
preliminary design and then cross check using the FE
option later.
Orion Training
Thank You

We hope you have found this


interesting and informative.

You might also like