You are on page 1of 16

Strand7

www.strand7.com 1

News.St7
Newsletter for Strand7 and Straus7 users
Strand7 is marketed as Straus7 in continental Europe

Strand7 R2.4 Released
As most of our users are aware, Strand7 R2.4 was released on December
7
th
2009. We are very grateful to our users who participated in the
Strand7 R2.4 Beta Program during 2009; their invaluable feedback
helped us achieve this goal.
In this issue of News.St7 you will find detailed articles on the new
concrete creep material model, the multi-point link and the new
geometry tools. We also illustrate the new friction options of the Point-
Contact element via a benchmark. Future editions of News.St7 will
cover more of the new features in Strand7 R2.4, but if you have a
suggestion about what you would like to see next, please let us know.
As always, this issue of News.St7 contains information on training and
exhibitions together with new Did You Know items.
If you have any feedback or suggestions regarding the content of News.St7 or you would like to
contribute a case study showing how you use Strand7 then please email news.st7@strand7.com. If
you would like to receive your copy of News.St7 automatically by email, simply send a blank email to newsletter-subscribe@strand7.com. All care is taken
to ensure that information in News.St7 is accurate and up to date at the time of publishing. However Strand7 Pty Ltd accepts no responsibility for
inaccuracies in, or changes to, such information.
Some of the new features in Strand7 were mentioned in
Issue 6 of News.st7. A few more features have been
added in the past year and here are some of the
highlights.
One important development is the new set of geometry
tools including the ability to create geometry within
Strand7. More about these later.
We have also added new material types including the
ability to use soil materials in brick elements. There are
now several new soil material models including Modified
Cam-Clay Soil, Mohr-Coulomb Soil, Drucker-Prager Soil,
and Linear Elastic Soil. The Cam-Clay Soil material is a
totally new development available for plane strain,
axisymmetric and 3D analyses. The Mohr-Coulomb and
Drucker-Prager Soils are extensions to the previously
available Mohr-Coulomb and Drucker-Prager elasto-
plastic material models, but have been enhanced with the
ability to consider fluid (pore) pressure and in-situ stress,
and to provide results as effective rather than total stress
(effective stress = total stress minus pore pressure). The
Linear Elastic soil material also considers pore pressure
and in-situ stress, and produces results as effective stress,
but it remains a linear elastic material.
The new Fluid material model is available for 3D brick
and 2D plane strain/axisymmetric elements. It can be
used for applications such as fluid-filled tanks and other
containers.
Creep of metallic and non-metallic materials can now be
simulated. A special feature on this topic, namely creep
and shrinkage of concrete, is included in this issue of
News.St7.
Another major development has been the enhancement of
the Strand7 Application Programming Interface (API). We
have added hundreds of new functions allowing users to
customise and automate modelling and post-processing
procedures, including the ability to embed a Strand7
graphical model window directly inside another
application. A summary of the improvements can be
found in this issue of News.St7.
In this Issue
- Strand7 R2.4 Features 1
- Multi-Point Link 2
- Concrete Creep and Shrinkage 6
- Modelling Tip - Concrete
Reinforcement 11
- API Functions 12
- Geometry Tools 12
- Benchmark 15
- Training and Exhibitions 16
Strand7 R2.4 Features
Issue 7, February 2010


Making Finite Element Analysis Easier
www.strand7.com 2
New Units have been added to Strand7, including the
calorie energy unit, stress in ksi, and time units for creep
material data and for use in the transient solvers
(milliseconds, seconds, minutes, hours and days). Scaling
of units now works on all data including Tables and
anisotropic material properties. In addition, changes to
the units no longer reset the Undo list, which means that
you can now Undo beyond a units change.
New display options
including Ignore
Cavity Loop on Face
free edge display
helps you identify
your models
unintentional free
edges.
The behaviour of
Solid View Dynamic Rotation has been improved to
maintain frame rate whilst moving the display.
There have also been many improvements to the way
results are displayed. Free body diagrams on any part of a
model can now be plotted using Element Node Reaction
Forces; you can also contour the reaction forces to help
you diagnose the
cause of non-
convergence in
Nonlinear
analyses. Contact
pressure at Face
Supports can be
plotted directly,
without the need
to extrapolate
element stresses.
This is particularly
useful in
nonlinear analysis
when the face
supports are of
the Compression-
Only type: a
contour of the
Face Support
stress clearly
shows the section
of the model that
is in contact.
New element attributes include the ability to define
individual node initial velocities for transient dynamic
analyses. Accelerations can now also be applied to the
nodes to generate inertial loads that are completely
arbitrary; this augments the previously available linear
and angular accelerations, which generate the same
accelerations over the whole mesh. Non-structural
masses can be offset from the node or element to which
they have been applied, taking into account the moments
generated by the acceleration field without the need for
rigid links.
Full documentation of the new features and changes to
old ones are included in the Strand7 R2.4 installation.
The files can be found in the Whats New directory:
Strand7 R24 for R23 Users.pdf and Whats New in
Strand7 R24.pdf (they are also available for download at
www.strand7.com)


One of the new link types introduced in Strand7 R2.4
is the multi-point link. There are two versions of this link:
1. User-Defined Multi-Point Link;
2. Interpolated Multi-Point Link.
The User-Defined version can be used to generate
arbitrary equations that constrain any number of degrees
of freedom to behave in a user-specified way. For
example, the equation DX(2)=0.5*DX(4) + 0.5*DX(6)
constrains DX at node 2 to be the average of DX at nodes
4 and 6. Here the user is free to specify all the degrees of
freedom and constants that make up the equation.
The Interpolated version requires a slightly different set of
inputs. These are:
1. The nodes to be constrained - one of these nodes
is designated as the slave node whilst all the
others in the link are designated as master nodes,
or the cluster of master nodes.
2. Whether the link is to act on the translational
degrees of freedom, the rotational degrees of
freedom or both translational and rotational
degrees of freedom.
With these inputs, the equations that relate the master
degrees of freedom (DoF) to the slave DoF are
automatically generated by Strand7. To generate these
equations, the link seeks to apply the average motion of
the cluster of master nodes to the slave node; the average
translation and average rigid body rotation of the cluster
are then applied at the slave node. To achieve this, a
least squares formulation is used based on the following
characteristics:
1. The translational DoF at the slave node are a
function of only the translational DoF at the
master nodes.
2. The rotational DoF at the slave node are a
function of both the translational DoF at the
master nodes and the geometry of the link (i.e.
the coordinates of the master nodes); there is no
direct coupling of the rotational DoF at the slave
node with the rotational DoF at the master nodes.
The general form for the system of equations is given by:

| |
(
(
(
(

=
(
(
(
(

N
N i
N
x
x
x
RHS RHS RHS x
LHS
LHS
LHS

2
1
2 1
2
1

i.e. Ry Lx =
i

Did You Know?
Numbered Backups
In R2.4 you can set the file
preferences to allow Strand7 to save
your model with a different numbered
backup every time you click save.
This should give users peace of
mind, knowing that model changes
can be retraced and, more
importantly, recovered.

Multi-Point Link


Making Finite Element Analysis Easier
www.strand7.com 3
where
i
x is the vector of slave DoF and
N , 2 , 1
x are the
vectors of master DoF. This over-specified system can be
solved via a least squares approach to give the slave DoF:
Ry L Lx L
T T
=
i

( ) Ry L L L x
T T
1
=
i

Depending on the link option Translations, Rotations or
Both, the
i
x vector will contain translational and/or
rotational degrees of freedom.
This link can be used to distribute load to multiple nodes
without influencing the stiffness of the structure.
Essentially it couples multiple nodes to a single slave
node without tying them all to each other only to the
slave node. Attributes applied to the single slave node are
distributed to the master nodes. This is useful if you want
to smear a mass onto a system, or apply load to many
locations from a single point. Using rigid links to do this
is not always desirable as it adds rigid bodies to your
model with infinite stiffness that can significantly
influence results, particularly local results. You can think
of the interpolated multi-point link as a soft version of
the rigid link.
Example: Mass Distribution in a Rocket
The simple rocket model shown
in Figure 2.1 uses links in this
way. We will explore the use
of rigid links and multi-point
links in various sections of the
rocket:
1) Nosecone
2) Instrument bay
3) Fuel tank
4) Burner
5) Tail and fins
The instrument bay has a 100kg
mass representing experiments,
electronics and payload. The
effect of this mass is distributed
with rigid links to 8 lugs.
The fuel tank has a 200kg mass
representing fuel. The inertia of
the fuel is distributed to the
walls of the tank by multi-point
links.
A 20kg mass in the tail
representing fin actuator
hardware, is also distributed
using multi-point links.

Figure 2.1: Simple rocket.
The rocket is 0.4m diameter and 3m long. It weights
48kg and carries a payload of 320kg. It is subject to a 6g
lateral acceleration in a tumbling re-entry as it slams back
through the upper atmosphere. A normal pressure
sufficient to produce this acceleration is applied to the
underside of the rocket, shown in Figure 2.2. The
freedom case type is set to Free Body Inertia Relief, so it is
not necessary to apply restraints to the model.
Each of the three applied nodal masses model different
payload behaviour, so the link type we choose is
important to model these differences realistically. The
100kg mass in the instrument bay represents a loaded
instrument tray, bolted to 8 lugs which are welded to the
walls of the rocket. This tray would add some stiffness to
the structure; thus, using rigid links to distribute the
inertial effects of the instruments is appropriate but comes
with the assumption that the tray is much stiffer than the
rocket walls and lugs. Alternatively the tray could be
explicitly modelled using shell elements with multi-point
links distributing the mass to the tray. The rigid links
attach to multi-point links in the lug holes to model some
bolted joint compliance.

Figure 2.2: Rocket re-entry pressure profile.
The fuel tank has a 200kg mass distributed to the walls
with multi-point links. This allows the walls to flex and
distort while still transferring the inertial loading from the
fuel. Had we used rigid links, the entire fuel tank would
behave as a rigid body, giving no stress results and
causing stress concentrations at the interface with the
other sections of the rocket. Figure 2.3 shows the links
representative of the instrument tray. Figure 2.4 shows
the fuel tank mass and links.

Figure 2.3: Instrument bay cutaway - mass and links.
The rockets tail contains some fin actuator hardware.
Each fin is independent and has a 5kg actuator. This is
modelled as a 20kg mass distributed to all four fins with
multi-point links. Had we used rigid links instead, the fins
would behave as though they were all connected by a
rigid body within the tail, which would not be reasonable.


Making Finite Element Analysis Easier
www.strand7.com 4
You could also choose to model the four actuators
independently, but having a single point mass to modify
in the future can be more efficient from a modelling
standpoint. Figure 2.5 shows the fin actuator mass
connected to the fins.

Figure 2.4: Fuel tank cutaway mass and multi-point links.

Figure 2.5: Rocket tail cutaway mass and multi-point links.
Examining results, the rocket bends about the 200kg fuel
load, which is the primary source of inertial loading
(11.6kN at 6g).

Figure 2.6: Rocket lateral 6g deflected shape and stress.
The six lugs attached to the bottom skin of the rocket
distort the skin surface causing local stress concentrations.
Although the tray is modelled with rigid links, sufficient
compliance is modelled in the lugs to maintain realism.

Figure 2.7: Instrument bay cutaway showing stress.


Figure 2.8: External view instrument bay deflection (100x).
The fuel tank section has the highest moment due to high
inertial loading there. Figure 2.9 shows the tanks internal
structure stress at 6g.

Figure 2.9: Fuel tank cutaway showing internal walls.


Figure 2.10: Fuel tank stress using rigid links.
Load Direction


Making Finite Element Analysis Easier
www.strand7.com 5
If rigid links were used to distribute the inertial load, the
entire tank would behave as a rigid body and there would
be no stress or distortion, as shown in Figure 2.10.
Interpolated multi-point links allow the natural distortion
and stress field to develop.
No significant stress or deflection occurred in the tail, and
the small 20kg mass was distributed to the four fins
without adding stiffness.

Figure 2.11: Tail cutaway stresses.
Example: Load Transfer at a Sliding Strap
The interpolated multi-point link can also be used to
model interaction between two structures, such as a
floating pier with a pile shown in Figure 2.12.

Figure 2.12: Floating pier showing tidal motion.


Figure 2.13: Pier model layout.
In this scenario, the pier is attached to the pile with a
loose fitting sliding strap. This strap allows the hollow
steel pile to deform within its grasp. The interpolated
multi-point link is well suited to this type of soft coupling,
where the strap can load the pile but doesnt add any
stiffness to it.
The pier and pile will be modelled using shell elements.
The strap is modelled as a beam element attached to a
multi-point link on the pile.


Figure 2.14: Strap modelling details.
The piles are fixed at the base, with two
of the three piles also restrained above
the floating pier to represent their
attachment to a non-floating pier. A
critical load case is an incident wave or
surge hitting the side of the pier. This is
modelled as a shell edge normal
pressure, and loads the pile through the
strap.
The pier was analysed using the linear
static solver. One column was critical
in a three-point bend mode with
ovalisation shown in Figure 2.15.
This ovalisation would not be allowed had we used rigid
links to represent the strap/pile interaction. The
interpolated multi-point link allows load transfer from the
pier to the pile and into the upper support without adding
stiffness to the pile, giving a more realistic response.


Figure 2.15: Pier stress and deflection (100x).
Strap
Floating Pier
Pile
Surge
T
i
d
e
s



Making Finite Element Analysis Easier
www.strand7.com 6
Interpolated multi-point links are a good tool for smearing
load or mass and for coupling structures in which one
member transfers load but not stiffness to the other.
The rocket and pier examples demonstrate the general use
of the interpolated multi-point links. This new capability
in Strand7 R2.4 expands the possibility for model
simplification and efficient model construction.
Interpolated multi-point links can be generated in a
number of ways in Strand7. One approach is to use
Create/Link from the main menu: select Multi-Point from
the dropdown list and then set the parameters (such as the
number of points to be linked and the degrees of freedom
to be coupled). Then click all the nodes in the cluster,
including the slave node. This approach is useful when a
relatively small number of nodes needs to be linked. For
linking larger clusters, use Tools/Auto Assign/Restraints.
With this tool you can select all the nodes in the cluster
by any convenient selection tool and then automatically
create a link to any desired slave node. There is even the
option of automatically creating a new node at the
average coordinates of the cluster and assigning it as the
slave node
Did You Know?
Select Brick or Plate Faces
In R2.4 you can select brick or plate elements based upon
the relative angles between free faces. This is useful for
situations where a surface pressure needs to be applied,
but the shape does not lend itself to easy selection using
Select by Region or any other convenient method.


With the release of Strand7 R2.4 comes a new
capability that allows users to model the creep and
shrinkage of concrete. These time-dependent and
inelastic deformations play a significant role in the long
term performance of concrete structures, and can be
factors of economic importance.
In practice when design engineers are investigating the
long term behaviour of concrete, they must generally refer
to one of many proposed design codes to establish
appropriate time-dependent properties. Among the most
commonly used concrete design codes are the European
code CEB-FIP Model Code 1990 [1] and the American
code ACI 209R-92 [2].
In this article we discuss:
1. extracting time-dependent material data from
these two particular codes; and
2. making the appropriate entries into Strand7 R2.4
in order to realise the various recommendations
within these design codes.
We will also examine the long term structural behaviour
of a reinforced concrete beam to further highlight the new
concrete creep and shrinkage modelling capabilities
within Strand7 R2.4.
Entering Design Code Data for Time-Dependent
Concrete Properties
The concrete design codes, CEB-FIP Model Code 1990
and ACI 209R-92, provide recommendations on the
following time-dependent concrete properties:
1. The development of modulus of elasticity with
time;
2. Creep; and
3. Shrinkage.
In this section we will look at some of the details of these
recommendations and at the same time highlight the
corresponding inputs required in the Strand7 R2.4
interface in order to comply with these concrete design
codes. (See the Online Help for a more complete
discussion on this subject.)
Development of Modulus of Elasticity with Time
It is well known that as concrete ages it increases in
strength and stiffness. The equations proposed by CEB-
FIP Model Code 1990 and ACI 209R-92 to model the
increasing modulus of elasticity with time are given in
Table 3.1.
To input these suggested equations for the time variation
of the elastic modulus in Strand7, choose Tables/Factor vs
Time and use the Equation Editor . The 28 day
modulus
28
E needs to be entered in the Structural tab of
the Element Property dialog box and the generated Factor
vs Time table needs to be assigned to the Modulus vs
Time dropdown in the Table/Time tabs of the Element
Property dialog box.
CEB-FIP Model Code 1990 ACI 209R-92
( )
5 . 0
28
28
1 exp
|
|
.
|

\
|
|
|
.
|

\
|
|
|
.
|

\
|
=
t
s E t E (1)

where
28
E is the modulus at
28 days and s is a constant
based on the cement type.

( )
28
E
t
t
t E
| o +
=
(2)
28
5 . 1
28
043 . 0
c
f E = (3)
where
28
E is the modulus at
28 days,
28 c
f is the 28 day
compressive strength; and
are constants based on curing
conditions and cement type.
Table 3.1: Concrete Design Code expressions for Modulus vs Time.
Concrete Creep and Shrinkage


Making Finite Element Analysis Easier
www.strand7.com 7
Creep
Creep is time dependent deformation that occurs due to
loading. The majority of design codes, including CEB-FIP
Model Code 1990 and ACI 209R-92, characterise creep
behaviour through the use of a dimensionless quantity
known as the creep coefficient ) , ( t t at time, t , and age
of first loading, t . This quantity is defined as the ratio of
creep strain to elastic strain:

( )
( )
( ) t c
t c
t
e
c
t
t
,
strain elastic
strain creep
, = =
(4)
As t approaches infinity, it is generally assumed that the
creep coefficient reaches an ultimate limit value ( ) t
u

where

( ) ( )
( )
( ) t c
t c
t t
e
u
c
u
= = ,
(5)
This final or ultimate creep coefficient is a useful and
commonly used measure of the capacity of concrete to
creep. The CEB-FIP Model Code 1990 and ACI 209R-92
expressions for the concrete creep coefficient are given in
Table 3.2.
CEB-FIP Model Code 1990 ACI 209R-92
Creep Coefficient:
( )
( )
( )
u
H
t
t
t
t |
t
t
3 . 0
,
|
|
.
|

\
|
+

=

(6)

Ultimate Creep Coefficient:
( ) ( )
cm RH u
f | t | =
(7)

Creep Strain:
|
|
.
|

\
|
=
28
) , (
E
t
creep
t
o c

(8)

Creep Coefficient:

( )
( )
( )
u
t
t
t
t
t
t
60 . 0
60 . 0
10
,
+

=

(9)

Ultimate Creep Coefficient:
( )
6 5 4 3 2 1
35 . 2 t =
u

(10)

Creep Strain:
( )
|
|
.
|

\
|
=
t
t
o c
E
t
creep
) , (

(11)

Table 3.2: Concrete Design Code expressions for the development of
concrete creep strain.
By examining the proposed equations for the concrete
coefficient in Table 3.2 (see Equations 6 and 9), it can be
seen that both the CEB-FIP Model Code 1990 and ACI
209R-92 code subscribe to the same type of hyperbolic
expression. In order to support both design codes,
Strand7 offers the following generalised hyperbolic law
*
:

( )
( )
( )
u
t
t
t
t |
t
t
o
o
o
|
|
.
|

\
|
+

= ,
(12)
To access this creep law in the Strand7 interface, you will
need to bring up the Creep Law Definitions dialog under
the Property menu (choose Property/Creep). By clicking
on the Creep Law dropdown list within this dialog, you
can access all of the various creep laws supported by
Strand7 including the Concrete Creep and Shrinkage -
Hyperbolic Law option.

*
The User Defined option can be used for any design codes which do
not subscribe to a hyperbolic law.

Figure 3.1: Creep Law Definitions dialog, Creep tab.
So how are the actual numerical values for the
coefficients in the generalised hyperbolic expression
determined? Well, the various parameters in the design
code equations (i.e.
H
|
in Equation 6;
RH
and
( )
cm
f |
in
Equation 7;
2
to
6
in Equation 10 etc.) are empirical
parameters that account for a variety of physical factors
including the composition of the concrete material, curing
methods, the humidity and the size of the concrete
member. Once you have quantified these physical factors
as they relate to the actual structure being analysed, it is a
simple matter of consulting the design codes to extract the
parameter values. While the majority of the parameters
extracted from the design codes are then easily matched
to the coefficients of the generalised hyperbolic law
offered in Strand7, it is worthwhile highlighting two subtle
aspects of this discussion.
Firstly, in order to account for the time variation of the
ultimate concrete creep coefficient as suggested by the
design codes - see the first terms in Equation 7 (i.e. ( ) t | )
and Equation 10 (i.e. ( ) t
1
) - an appropriate Factor vs
Time table may be assigned. Equations 13 and 14 in
Table 3.3 show the actual design code recommendations
for the influence of time on the ultimate creep coefficient.
We suggest using the Equation Editor within the Factor vs
Time table to generate data according to these equations.
The second issue that prompts further consideration
relates to the code definitions of creep strain given in
Equations 8 and 11. According to the CEB-FIP Model
Code 1990, calculating the creep strain from creep
coefficient involves a division by a constant value of the
elastic modulus
28
E (see Equation 8). To specify this
aspect of the concrete creep model within Strand7, the
Constant Modulus option within the Creep Law
Definitions dialog (see Figure 3.1) needs to be set.
Conversely, the ACI 209R-92 model requires that the
creep strain be determined by dividing by an elastic
modulus that varies with time (see Equation 11). For this
situation, the Constant Modulus option needs to be
cleared.


Making Finite Element Analysis Easier
www.strand7.com 8
Strand7
Parameter
CEB-FIP Model
Code 1990
ACI 209R-92
o 1.0 0.6
|
Determined from
code based on the
relative humidity
and size of member
10
o
0.3 1.0
u
| ( )
cm RH
f |
determined from the
relative humidity,
size of member and
compressive
strength
6 5 4 3 2
35 . 2 =
u
determined from relative
humidity, size of
member, curing
conditions and concrete
composition
u
| Factor
vs Time
Table
( )
2 . 0
1 . 0
1
+
= t |
(13)
118 . 0
1
25 . 1

= t
(moist cured) (14a)
094 . 0
1
13 . 1

= t
(steam cured) (14b)
Constant
Modulus
Checked Unchecked
Table 3.3: Summary of Strand7 inputs for matching design code
recommendations for creep.
Shrinkage
Shrinkage is the time-dependent deformation that occurs
independently of any applied loading. The recommended
expressions for the development of shrinkage strain over
time provided by the CEB-FIP Model Code 1990 and ACI
209R-92 codes are given in Table 3.4.
CEB-FIP Model Code 1990 ACI 209R-92
( )
( ) ( )
u sh sh
t h h
t
,
5 . 0
2
0
350
c
t
t
c
|
|
.
|

\
|
+

=

(15)

( )
RH cm s u sh
f | c c =
,

(16)

( )
( )
u sh sh
t
t
,
c
t |
t
c
+

=
(17)
where 35 = | (moist cured)
or 55 = | (steam cured)
7 6 5 4 3 2 1
6
,
10 780
sh sh sh sh sh sh sh
u sh

c =


(18)

Table 3.4: Concrete Design Code expressions for the development of
concrete shrinkage strain.
Once again, in order to accommodate both design
recommendations, a generalised hyperbolic expression
for the Shrinkage Strain is offered in Strand7:

( )
( )
( )
0
,
s
s
s
s
s
s
t
t
t c
t |
t
t c
o
o
o
|
|
.
|

\
|
+

=
(19)
To access this hyperbolic law for shrinkage in the Strand7
interface, click the Shrinkage tab of the Creep Law
Definitions dialog and choose the Shrinkage Formula
option.

Figure 3.2: Creep Law Definitions dialog, Shrinkage tab.
The process of determining the numerical values for the
various parameters for shrinkage hyperbolic law follows
the same steps discussed for the creep parameters. Again,
you will need to consult the design codes to determine
which physical factors influence the shrinkage properties
(e.g. the relative humidity, the size of the member, the
cement type, the compressive strength) and decide on the
appropriate values of these factors as they apply to your
analysis. Following this, it is a straightforward process of
obtaining the required parameter values for the
hyperbolic expression and matching the inputs into
Strand7. A summary of the entries required in Strand7 to
match the design code recommendations for shrinkage is
provided in Table 3.5.
Strand7
Parameter
CEB-FIP Model
Code 1990
ACI 209R-92
s
o
1.0 1.0
s
|
( )
2
350
o
h h
where h and
o
h are measures
of the size of the
concrete
member.
35 = | (moist cured ) or
55 = | (steam cured)
s
o
0.5 1.0
0 s
c ( )
RH cm s
f | c
determined from
concrete
strength, cement
type and relative
humidity
7 6 5 4 3 2 1
6
10 780
sh sh sh sh sh sh sh



determined from relative
humidity, size of member,
curing conditions and
concrete composition.
Table 3.5: Summary of Strand7 inputs for matching design code
recommendations for shrinkage.


Making Finite Element Analysis Easier
www.strand7.com 9
Example: Creep and Shrinkage of a Reinforced Concrete
Beam
Now that you know how to extract and input code based
concrete creep and shrinkage properties into Strand7, lets
have a look at an actual analysis. This example
investigates the effects of concrete creep and shrinkage on
a pre-tensioned reinforced concrete beam over a duration
of 50 years. The concrete beam was modelled using
brick elements and the reinforcement modelled using
truss elements assigned to various string groups. The
concrete beam was fixed at both ends and the structural
response was compared under two different loading
scenarios:
1. with no loading; and
2. with an applied pre-tension in the top layer of
reinforcement within the flange.
Before investigating the results of the analysis, we will
have a look at some specific details of the Strand7 model
setup that are particularly relevant to performing a
concrete creep and shrinkage analysis in Strand7.

Figure 3.3: Sketch of the reinforced concrete beam. The beam outline is
drawn in black and is meshed using brick elements. The blue lines
represent the pre-tensioned bars, while the red line represents the
reinforcement bar. These are meshed with beam elements.
Firstly, lets review the process of assigning concrete
creep and shrinkage properties in the model. From the
main menu Property/Creep was selected and from the
Creep Law dropdown list, the Concrete Creep and
Shrinkage Hyperbolic Law was accessed. Defining the
required coefficients could be performed by following the
process described previously, but in this case Strand7
R2.4 includes creep and shrinkage data for concrete in the
library. By clicking the Import Data button and
selecting ACI-209 R-92 Concrete Creep +
Shrinkage/Steam Cured Concrete the appropriate
coefficients were filled in. Users have the option of
adding to the library (click Export Data ), so if
concrete creep and shrinkage is a consideration that you
will need to tackle often, we suggest that you make use of
this facility.
Next we assigned the concrete creep and shrinkage data
set to the appropriate brick property. This was done by
clicking Property/Brick, choosing the Nonlinear tab and
then selecting from the dropdown list under Creep Data.
All the creep sets that are generated under
Property/Creep are made available here to be selected as
part of the Element Property definition (Figure 3.6).

Figure 3.4: Import Creep Data dialog, showing the available creep data.

Figure 3.5: Creep Law Definitions dialog after the ACI 209 data is
imported.

Figure 3.6: Selecting creep data in material property dialog.
As well as defining the creep and shrinkage properties, we
needed to assign the elastic properties for the concrete
material. To comply with the ACI 209R-92 code for the
development of the modulus of elasticity with time (see
Table 3.1), the following steps were undertaken:
1. From the main menu Table/Factor vs Time was
selected and the Equation Editor was used to
enter the data in Figure 3.7 (this is Equation 2
without
28
E where o =1.0 and | =0.95);
Brick mesh
Reinforcement bar
Pre-tensioned bars


Making Finite Element Analysis Easier
www.strand7.com 10
2. The 28-day modulus (
28
E ) was assigned by
clicking Property/Brick and entering the
appropriate value in the Structural tab;
3. Within the Brick Element Property dialog, the
Tables tab and Time tab were selected and the
Factor vs Time table was assigned in the Modulus
vs Time dropdown.

Figure 3.7: Strand7 Equation Editor for Factor vs Time tables.
Once the mesh setup was complete and all the material
properties were assigned, we solved the model using the
new quasi static solver (Solver/Quasi Static) with the
Creep option set. This solver allows you to calculate the
time-dependent response of a structure subject to any
loading conditions and initial conditions, but ignores any
dynamic (inertial) effects. It is, therefore, the most
suitable solver for performing a creep analysis. Also, by
using the time step editor in the quasi static solver (Time
Steps) we were able to specify an appropriate time
stepping scheme over the 50 year period of interest.

Figure 3.8: Quasi Static Analysis setup. Note that the Nonlinear Material
option is enabled to account for the Modulus vs Time table.
As the model is relatively large and the mesh is made up
of predominantly brick elements, the newly available
iterative PCG scheme was used (see Scheme/Iterative
(PCG) in the Quasi Static Solver dialog). This scheme is
most useful for meshes that have a large stiffness matrix
with predominantly brick elements, and in general has
shorter solver running time compared to the Sparse
scheme.
After solving the model we were able to investigate the
structural response of the reinforced beam over time as it
experienced creep and shrinkage deformations. As
mentioned previously, analyses were performed with and
without pre-tension. A comparison of these two cases is
shown in Figure 3.9 where the evolution of the concrete
fibre stress is plotted.

Brick Stress vs Time
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
0 5000 10000 15000 20000
Loading Time (Days)
S
t
r
e
s
s

(
M
P
a
)
No Preload
Preload

Figure 3.9: Brick ZZ stress vs time plots for the beam without and with
pre-tension.

Did You Know?
Plate and Brick User Contour
For plate and brick elements a User Contour based on stress
and strain result quantities can be defined in Results Settings.
This is useful in plotting the contours of derived quantities
based on the elemental stress and/or strain, for example a
material failure criterion. This is also available for the Peek
tool and Results Listings function.
In this example, we derive and contour a concrete cracking
index according to
( )
( ) t
t f
I
t
cr
1
o
= , where ( ) t f
t
is the tensile
strength at age t obtained from concrete design codes and
( ) t
1
o is the maximum principal stress calculated by Strand7.
Since only the tensile stress is required, the function IFPOS
can be used in this case, as shown below (IFPOS(x) returns x
when x is positive and zero otherwise). The contour below,
obtained with the equation 0.35*SQRT(32)/IFPOS([S11]),
indicates that some cracking can be expected in the concrete
beam due to the creep and shrinkage the results shown are
for a case where there is no pre-tensioning.





Making Finite Element Analysis Easier
www.strand7.com 11
For the first case where no loading is applied, it is no
surprise that the instantaneous response (at time = 0) is a
stress state of zero. However as time continues, the
concrete material undergoes compressive shrinkage
deformation. A consequence of this deformation, along
with the beams end fixity, is that the concrete beam is
placed in tension. The creep deformations within the
concrete would perhaps have a relieving effect however
the resulting tensile stresses are not desirable and may
lead to cracking.
To mitigate the possibility of long term cracking, pre-
tension can be applied to the reinforcement. From Figure
3.9, it can be seen that pre-tensioning the reinforcement
bars places the concrete beam into compression. While
the creep and shrinkage mechanisms influence the long-
term stress field within the concrete material, the desired
compressive state is maintained.
The model used in this article can be downloaded from
www.strand7.com/news.st7.htm
References
[1] CEB-FIP Model Code 1990, Comit Euro-International
du Bton, 1990.
[2] ACI Committee 209, Prediction of Creep, Shrinkage
and Temperature Effects in Concrete Structures, ACI
209R-92.




One of the common questions we receive on concrete structures is related to the modelling of concrete reinforcement. As with other
structural modelling it depends on the level of accuracy and local details required by the user. In this issue, some of the common
approaches are presented in the order of increasing complexity.
1. Beams Model 2. Plates and Beams
Model
3. Simple Bricks and
Beams Model
4. Detailed Bricks and
Beams Model

This option is used if the global
response of a beam or column-
like structure is of interest. The
concrete and reinforcement are
modelled as coincident beams
of different properties and the
reinforcement is positioned in
the correct location using the
Offset Attribute.

Pros:
- Very simple to model.
- Good global representation.
- Solution time is very short.
- Can generate bending
moment and other quantities
conveniently.


Cons:
- Limited to beam and column
type structures.
- Can be difficult to model for
complex reinforcement.
This option is used if the
structure of interest has slabs
or walls in addition to beams
and columns. The
reinforcement can be modelled
either as beams located at the
correct locations, or as plates
using an equivalent smeared
representation.

Pros:
- Good representation of a
general structure.
- Solution time is relatively
short.




Cons:
- The interaction between
concrete and reinforcement
is difficult to model.
- Can be difficult to model for
complex reinforcement.
This option is used if a
relatively detailed model is
required. The concrete is
modelled as brick elements
and the reinforcement as
beams. The beams are located
on the brick nodes.



Pros:
- Good representation for a
detailed model.
- Can model the interaction
between concrete and
reinforcement.



Cons:
- Solution time can be long.
- Stress singularity at the ends
of the reinforcement bars.
This option is a more detailed
version of Option 3. The
location and diameter of each
slot in concrete is modelled.
The reinforcement bars are
joined to the concrete via rigid
links and possibly contact
elements.


Pros:
- Most accurate
representation of all options
presented.
- Singularities are avoided.
- Can model the interaction
between concrete and
reinforcement.

Cons:
- Solution time is the longest
of all representations.
- The model can be time
consuming to create.

As you can see, the modelling options depend on the type of structure to be analysed and the results of interest. If you are interested
in global responses, Options 1 and 2 are more appropriate. Options 3 and 4 are appropriate for detail analysis. It is also possible to
create a model which features more than one option introduced above. For example, a complete building model can be constructed
using Option 2, with all columns modelled using Option 1. We also encourage you to seek out modelling methods that are not
covered here, but can give a good representation of the structure
Modelling Tip Concrete Reinforcement


Making Finite Element Analysis Easier
www.strand7.com 12
The Strand7 Application Programming Interface (API)
received major upgrades for R2.4, including access to
hundreds of new functions. New capabilities found in
Strand7 R2.4 are included in the API, allowing
construction sequences, moving loads, import and export
of geometry files and other formats, geometry cleaning
and tools, and even automeshing.
This suite of tools gives you access to the full flow of
model creation and analysis, from geometry import,
through automeshing and application of attributes,
solving, and on
to results post-
processing all
from within the
Strand7 API
environment.
New API
functionality allows
you to open interactive
Strand7 model view windows,
giving instant graphical feedback
including results. These windows can also be embedded
directly within your Strand7 API program for an integrated
appearance. You can create and display animated results,
and custom result files can be built up within the API, for
example fatigue life or combined damage contours.



The ability to launch multiple Strand7 solvers at once
allows for parallelisation of large batch jobs in todays
multiprocessor desktop environments. This can
dramatically decrease your time to results for a design-of-
experiments or optimisation approach to design; even
with a single Strand7 licence you can launch multiple
solvers on the same PC.
The Strand7 API interfaces with many different
programming environments and languages, including
C
++
, Delphi, Intel Fortran, MS Visual Basic, MS Excel
VBA, MS Visual C
++
and, new for R2.4, MS Visual
C
++
.NET and Matlab 7.3

Working with geometry entities has become easier in
Strand7 R2.4 with the new Geometry Tools. For
example, a geometry face can now be split by vertices or
by a plane entirely within Strand7, making the process of
generating a compatible mesh much simpler than before.
Some other great additions to the Geometry Tools include
grafting edges to faces, generating faces from beam and
plate elements, intersecting and morphing edges; all
accessible under the Tools/Geometry Tools submenu.
The following section highlights some of these modelling
techniques when working with geometry entities.

Face from Plate and Split Face by Vertices


Plate elements turned into
geometry faces.




Geometry faces split and
common edges created by
vertices.




Geometry faces can be generated within Strand7 by
converting plate elements into faces. For simple
API Functions
Geometry Tools


Making Finite Element Analysis Easier
www.strand7.com 13
geometries, this can be a quicker and easier method than
creating geometric entities and importing them from a
CAD model.
Should it be necessary, vertices can be created to define a
line for geometry faces to be split. A split line on a planar
face will remain straight while the split line on a curved
face will follow its face definition.
Rebuild Faces

Bad geometry definition
rebuilt to improve mesh
quality.




Sometimes geometry imported from CAD has poor
underlying B-spline surface definitions. Try rebuilding it
to refit a new, better quality surface over the existing
surface.
Graft Edges to Faces and Morph Edges


Curved edge grafted to
curved face.




Blue edge morphed
towards the red edge.




A mesh generated from a grafted face will inherit its
grafted feature, thus creating interfaces between the
geometric entities.
Edges at the interface can then be morphed together to
create proper intersections for compatible automeshing.
Did You Know?
Skip Transitioning
Skip Transitioning is an automeshing option that aims to
produce a mesh as regular as possible by ignoring small
features in the geometry. This option can be used in
conjunction with the Vertex Mesh Size attributes to create a
mesh with specific boundaries such that it can be directly
stitched onto the existing mesh.

Initial Mesh
A modification is required to include
an additional hole to the existing
mesh.

Using the new geometry tools such
as Face from Beam Polygon and
Graft Edges to Faces, a geometry
face with a cavity can be created.
Additional vertices with matching
Mesh Size attributes are created.


Modified Mesh
The geometry face is surface meshed
with the Skip Transitioning option
and the mesh will be modified with
the required feature.


Skip Transitioning can also be useful for skipping over small
features, such as small mismatched edges producing small
element edge lengths.

With
Transitioning


Skip
Transitioning




Making Finite Element Analysis Easier
www.strand7.com 14
Split Face by Plane


Geometry faces can also be split by plane. The cutting
plane is defined by a user cartesian coordinate system of
choice.
In addition to the splitting, geometry faces at the cutting
plane can be optionally generated. This is particularly
useful to separate the volumes for solid meshing later on.
Face from Beam Polygon





A closed polygon formed by beam elements can also be
used as a definition to create geometry faces.
An angle tolerance is available that allows you to control
whether curves should be smooth or faceted.
You can generate planar faces in this way, or you can
generate curved faces by projecting the polygon onto
another existing face.
In the example below, the magenta beam polygon is not
planar. If the Face from Beam Polygon tool is used
without choosing a Target Face, Strand7 calculates the
least-squares best fit plane and creates the face on that
plane (shown in green) with the vertex locations projected
perpendicular to the new plane. If the Target Face is
selected (here we have used the blue face), the resulting
face (the red face) is created using projection
perpendicular to that target face.

Intersect Edges





Edges can be easily intersected in a few clicks.
Intersections are detected if the distance between the
edges falls within the specified tolerance.
The faces will remain unconnected but an option to split
the faces while intersecting the edges is provided.
Splitting the faces will achieve mesh compatibility.


Making Finite Element Analysis Easier
www.strand7.com 15
Face from Cavity


Geometry face created
from cavity.



Filling in a cavity is easily achieved by generating a
geometric face from a cavity loop.
Faces created from cavities will inherit the surface
definition from the face with the cavity; in this instance it
is cylindrical
Did You Know?
Browse
If you are trying to find a model and cannot remember its
name you can try using File/Browse in Strand7.



Navigate to a likely model location in the Browse window
and click Browse. This will display an image of the last
saved model view with the file name of all Strand7 files
located in this folder. A little contour bar is displayed in the
bottom left hand corner of the image if the model has
results. Also, if you right-click a model image you can
open, rename, delete or see information about the file.


In this edition we consider a problem based upon one of
Strand7s Verification models, documented in the
Verification Manual (the Verification Manual is a 300+
page book, shipped in PDF format, and is located in the
Verification folder of your Strand7 R2.4 installation).
This problem makes use of the Plastic Friction Model,
another new addition to R2.4. The previously available
Elastic Friction Model assumes that the friction force due
to contact is equivalent to a nonlinear spring. This means
that if the movement direction reverses, the friction force
does not reverse until zero relative displacement is
reached. As a result, this friction model is only valid in
situations where the movement does not stop or reverse
direction.
The Plastic Friction Model accounts for changes in
direction, so if the direction of movement reverses the
frictional force reverses accordingly. This allows the
friction model to demonstrate hysteresis and should give
more realistic friction forces when contact is lost and
regained between entities.

Contact
Movement


Plastic
Friction
Force

Another improvement in the friction capabilities of the
new contact element is the addition of an elliptical yield
surface, in addition to the previously available rectangular
yield surface. In the rectangular option, equilibrium
satisfies the following equations independently:
N F
1 1
= and N F
2 2
= ,
where
1
F and
2
F are the frictional forces in the element's
1 and 2 principal directions respectively,
1
and
2
are
the corresponding coefficients of friction, and N is the
normal force in the element. With the elliptical option,
the equilibrium equation is
1
2
2
2
2
1
1
=
|
|
.
|

\
|
+
|
|
.
|

\
|
N
F
N
F

.
The model uses 20 2D plane stress elements to define the
block that is supported by 21 Normal Point Contact
elements. To improve the accuracy of the model it is
necessary to introduce damping locally to the contact in
the vertical direction. This has been achieved by using
Benchmark


Making Finite Element Analysis Easier
www.strand7.com 16
small Spring-Damper elements between the contact
elements and the block.

The nonlinear transient dynamic solver is used with the
nonlinear geometry option set. Viscous damping does
not need to be added in the solver because the viscous
damping properties of Spring-Damper elements are
always included in nonlinear transient analyses.
The initial velocity is applied as an initial condition within
the solver dialog (using VX=1) and is applied to all the
nodes in the model. It is only used on those nodes which
are not restrained and are therefore free to move in the
direction of the initial velocity.
The box slides with initial velocity of V0=1m/s along a
horizontal surface with friction coefficient =0.3. The
distance the box will take to slide to rest depends on the
energy balance of the system. By energy methods the
analytical result for the distance is:
= =
g
V
X
rest
2
2
0
0.16995m

A graph of the horizontal displacement with respect to
time is shown in Figure 4.2. It takes just over 0.3 seconds
for the box to stop moving and reaches a distance
0.17045m at the end of the analysis, 0.3% above the
analytical solution. The slightly higher value is caused by
the initial settling period where the contact has not yet
established itself fully giving a lower friction force for a
short time.

Figure 4.2: Horizontal displacement vs. time (s).
This model is available to download from the News.St7
page of our website www.strand7.com/news.st7.htm

The training calendar for the first half of 2010 is still
being finalised, but the following courses have been
scheduled:
Date Course Title Location
22-24 February Strand7 Essentials Perth, Australia
25 February Nonlinear Analysis
with Strand7
Perth, Australia
26 February Dynamic Analysis
with Strand7
Perth, Australia
1-3 March Strand7 Essentials Cambridgeshire,
UK
4 March Nonlinear Analysis
with Strand7
Cambridgeshire,
UK
5 March Dynamic Analysis
with Strand7
Cambridgeshire,
UK
To register, please go to www.strand7.com and follow the
links.
We are also planning a Strand7 Workshop for our UK
users on the 8
th
and 9
th
of March. This will be held in
Cambridgeshire - details on content will be released on
the Strand7 website shortly


Between now and end of May, we will be exhibiting
Strand7 at the following events: OTC 2010 (Offshore
Technology Conference) and NASCC 2010 (North
American Steel Construction Conference).
3-6 May 2010 OTC Houston, USA
12-14 May 2010 NASCC Orlando, USA
Other exhibitions will be announced soon on the Strand7
website and in the next edition of News.St7


Head Office
Strand7 Pty Ltd
Suite 1, Level 5
65 York Street
Sydney NSW 2000
AUSTRALIA
Tel +61 2 9264 2977
Fax +61 2 9264 2066
Email info@strand7.com
Web www.strand7.com

Training
Exhibitions

You might also like