Professional Documents
Culture Documents
NQ 49840
NQ 49840
By :
Evan C. Bentz
O Evan Bentz
2000
National Libraiy Bibliothque nationale
du Canada
Acquisitions and Acquisitions et
Bibliographic Services services bibliographiques
395 Wellington Street 395. rue WdlingtOn
OttawaON KlAON4 OttawaON K l A W
canada CaMda
The author has granted a non- L'auteur a accord une licence non
exclusive licence allowing the exclusive permettant la
National Library of Canada to Bibliothque nationale du Canada de
reproduce, loan, distribute or seli reproduire, prter, distribuer ou
copies of this thesis in microfom, vendre des copies de cette thse sous
paper or electronic formats. la forme de microfiche/film, de
reproduction sur papier ou sur format
lectronique.
Year: 2000
Four easy to use programs have been written that allow for state of the art
sectional analysis of reinforced concrete blocks, plates, beams, columns and shellS.
Unlike most sectional analysis programs, these programs include the effects of shear on
behaviour. They are based on the assumption that plane sections remain plane, that there
is no transverse clamping stress, and that the biaxial behaviour can be modelled wel! by
be reasonable.
The programs are fieely available on the World Wide Web at the listed addresses:
This thesis describes the MCFT in detail as implemented in the programs as well
as explaining new constitutive relations employed for the behaviour of concrete in
tension. The strongest feature of the new programs is the employment of the longitudinal
stiffness method, developed for this thesis, which calculates the shear stress profile for a
beam or shell much faster and with more numerical stability than the previous state of the
art.
The programs are verifted against a set of experiments as well as against two new
shear expenments perforrned for this thesis. They indicate the programs are good at
534 beams and shown to predict shear strengths with an average experimental over
predicted shear strengh ratio of 1.05 and with a coefficient of variation of 12%. This
compares favourably to the AC1 code prediction ratios that have an average of 1.20 and a
It is suggested that the programs in this thesis represent a good first step in
allowing rational, state of the art cornputer programs to be directly allowed in the code
Chapter 1: htroduction.................................................................................................... 1
1- 1 Analysis of Structures ............................................................................................ 1
1-2 Sectional-Based Analysis of Concrete Swctures ....................... . . . . ................ 3
1-3 Bnef Description of t he Programs ......................................................................... - 6
1-4 Types of Sectional Analyses................................................................................... 8
1-5 Assumptions in Sectional Analysis ....................................................................... -8
1-6 Contents of this Thesis ......................................................................................... 10
vii
List of Figures
1.1 Air Force warehouse failure in Shel by. Ohio ............................................................ - 2
1-2 Hanshin Expressway piers collapse in Kobe. Japan ................................................... 2
1-3 Detennining sectional forces using plane fiame analysis ........................................... 4
1-4 Determining sectional forces using finite element analysis ........................................ 5
1-5 Sectional analysis of Reinforced Concrete ................................................................ -9
1-6 Components of sectional analysis methods ............................................................... -9
1-7 Example of transverse stress in web of beam ........................................................... 11
During the l m twenty five years, a considerable arnount of research has been
conducted world-wide with the aim of developing behavioural modets for reinforced
concrete in shear comparable to the rationality and generality of the plane-sections theory
for flexure. This research is comprehensively reviewed in the Dec. 1998 state-of-the-art
report by ACI-ASCE Committee 445 "Shear and c ors ion''. One group of rationai
models for shear, developed at the University of Toronto, is known as the "Modified
Compression Field Theory" (MCFT)'. The programs in this thesis use this model.
The research described in this thesis was commenced in the belief that recent
advances in both computational power and behavioural understanding make possible the
development of a new generation of design models for reinforced concrete subjected to
shear, To this end, a series of four programs have k e n written for the shear analysis and
design of rein forced concrete elements. These prograrm incorporate a number of
significant advances in the shear analysis of reinforced concrete. In the author's opinion,
the most important attribute of these programs is that they are designed to help the user
understand the response of reinforced concrete elements loaded in shear.
These programs also are designed to provide only one answer for a given
problem. Some engineering analysis programs provide the ability to "tune" results by
changing one analysis parameter or another. While this flexibility is usefil, it can be
tempting to try to achieve unnaturally good agreement with experiments that have already
been performed. It is felt that this does not serve the engineering community well. By
calculating a single definitive shear suength, say, for a given set of basic material
properties and sectional geometry, it is felt that a more stringent set of rules for judging
quality is established,
Finally, these programs are designed to provide a numerical test-bed for large and
cornplex problems. It is known, for example, that larger lightly reinforced structures tend
to fail at lower shear stresses than smaller ones. Clearly, there i s a limit to how large an
experiment can be to test for this phenomenon. By making programs that are based on
rational models, it is possible to make the best use of the few large wncrete shear tests
that do exist. With a demonstrated ability to predict trends that include these large tests,
it becomes possible to predict, with reasonable confidence, shear response when
experiments are not practical.
In using the sectional analysis approach, the problem o f detennining the response
of a reinforced concrete structure to applied loads is broken up into two interrelated tasks.
First, the sectional forces at various locations in the stmcture caused by the applied loads
are determined. This step is usually performed assuming that the structure remains
linearly elastic. Then the response of a local section to the sectional forces is determined.
Tn this second step, which is the sectional analysis, the non-linear characteristics of
cracked reinforced concrete are taken into account. Two examples of calculating the
sectional forces in reinforced concrete stnictures are shown in Fig. 1-3 and Fig 1-4. For
the simple building fiame shown in Fig. 1-3, the axial load, N, the moment, M, and the
5
The contribution of this thesis to the field is that it provides immediately useftl
programs that can be used by engineers and researchers to perform non-linear sectional
analysis. These programs have been written so that they quickly allow checking o f input
and output data for errors. Additionally, in the process of viewing the results, the
engineer using the program cannot help but learn about the behaviour of reinforced
concrete. New analytical methods have been developed that improve the stability of the
analysis while also increasing the speed many times over the previously existing state-of-
t he-art proceciures.
"Although too complex for regular use in the design of simple beams, the
procedure has value in its ability to provide a rational method of analysis and design
for members having unusual or complex geometry or loading, or whenever a more
thorough analysis is warranted."
It is believed that the advances made in analysis techniques since that time and the
advances made in cornputer power now mean that Response-2000 can in fact be used in
day-to-day office practice.
The founh program, Shell-2000 perfonns analyses for plates and shells subjected
to al1 of the 8 force resultants shown in Fig 1-4 c. This new program performs a more
rigorous analysis for out of plane shear than previously available. This program can be
thought of as the n e a generation of program ~he11474'.
A user manual for these four programs is given in Appendix A of this thesis.
More importantly, each of the four programs is fieely available via the World Wide Web
at the following addresses:
3 0 Frame
2D Frarne 3D Sheli
-..ml - 0 1 1 1
20 Frame 3 0 Shell
Basis: Basis:
2D nodes in a Iine 3 0 nodes in a line
Finite Element Finite Element
TEMPEST RASP
Loadina
N M V
10 Frame 3 0 Block
Basis: Basis: Basis:
1 0 node at a point 2 0 node at a point 3 0 node at a point
Finite Element Finite Element '
None TR fX
Loading Loading Loading
N NxN,V* N x N v N z Vm V, V,
Membrane-2000
the load itselc which will tend to locally increase the strength. This is one reason that
short beams are noticeably stronger in shear than long beams with the same cross section.
Figure 1-7 shows the results of two non-linear finite element analyses performed
with program TRIX8,which is a membrane element non-linear finite element analysis
program based on the Modified Compression Field Theory. A 1.85 metre deep bulb-tee
type section was loaded with a central point load on a 10 meue span. The analyses were
performed to evaluate the appropriateness of the assumption of no clamping stresses in
the transverse direction. For the first analysis, the strands in the beam were not
prestressed, making the beam a reinforced concrete beam, while for the other, the strands
were stressed to 1 100 MPa ( 6 W of ultimate stress). The plots show the transverse
clamping stress at the top, middle and bottom of the web of the beams, al1 for a total
applied load level of 3000 kN. The middle region of each shear span is highlighted in the
diagram. It can be seen that in this middle region the effects of the transverse stresses
near the support and point load have largely dissipated. Note that for the reinforced
beam, there is little transverse stress in this middle region, but for the prestressed beam,
there is a more pronounced clamping at mid-depth. This implies that a sectional mode1 of
this prestressed beam would be more conservative than for the reinforced concrete beam
as the analysis would ignore the beneficial effect of this small compressive, clamping,
stress. Based on analyses such as these summarised in Fig. 1-7, it can be wncluded that
the traditional assumption that the transverse stress is negligible is a very good one.
/ -.dr
AL
B, /
C ,l
:
*
transverse stress
c
5-1-
-T
l S Y 8 2mmn
32 S13
lOGp-=
5.50 mmk
f 1500 kN 1500 k ~ f
ha+
Load
2 -r
4 Load
Trartswrse Stress at S e d o n 6 4
!
4 Load
I I 1 I
chapter is written in a "tutonal" style to allow the reader to follow dong in trying the
programs.
Chapter 3 explains the implementation of the MCFT in two dimensions and three
dimensions. Chapter 4 extends that with an explicit discussion of the crack check This
crack check is a necessary part of the MCFT to ensure that equilibrium can be maintained
at a crack.
Chapter 5 describes the concrete constitutive models used for the programs. New
relations for the tension strength of concrete and tension stiffening are presented. These
were found to be necessary to better capture observed behaviour of beams.
Chapter 6 describes the new method of calculating the shear stress profile in a
beam or shell. This is one of the most important parts of this thesis as it has allowed a
substantial increase in performance and stability over previous methods.
Chapter 7 provides brief descriptions of how each of the programs work. As the
programs contain a totd of about 150,000 lines of CH, the level of detail cannot be very
high for this chapter.
2- 1 Program Installation
To install the programs from this thesis, simply copy them into a new directory
and unzip the zip files. Consult your Microsoft Windows manual to find how to make a
shortcut to the program or to add them to the start menu.
p=$-
All dirneruioru in miHimetreg
Minimumc k r cover : 6 mm
a = 6mm w20
t=l.akp.(M)
By default, membrane-2000 starts with PV20 loaded, so to see the element after
starting the program, simply click on the cross section icon in the toolbar, which looks
like a little membrane element or select the menu option "View 1 Cross Section". The
figure shown above is a direct print o f the page that will appear.
The drawing attempts to document al1 the input parameters of the mode1 t o allow
for easy error checking or quick documentation o f a design. The properties shown o n the
page may be changed using the "define" menu. Additionally, double clicking on the
drawing itself allows easy access to the define menu. For example, t o change the stress-
strain properties of the reinforcement in the X direction, it is possible to go to the "Define
1 Materials" menu option, or simply t o double click on the drawing near the stress-strain
line o f the x-steel.
The screen will change to a 9-plot view as shown below. This is a standard view for the
programs explained in this thesis. Each plot represents one variable of the solution for
the panel PV20. For Membrane-2000, each plot is a fiill load-deformation plot. Some of
the experimental data fiom the test1' are included as well for cornparison. Note that the
while the experimental correlation is not especially good for some of the variables, the
overall behaviour as represented by the Shear-y, plot is quite reasonable.
Each of the programs in this thesis will work with either SI metric, US customary
units, or kg/cm2 units as used in, for example, Japan. By defaulf the programs start up in
SI metric (See Section 5-1 1 of the Appendix for information on how to change the
default start units). The units may be changed in the "Options 1 Preferences" menu. For
this example, stresses are in MPa, and strains are in parts per thousand (x 10'~or mmlm).
On the lefi of the screen is a "control plot." It has crosshairs showing the
currently selected load stage. This is the state that the crack diagram represents, with the
crack width shown in mm. The red vertical line on the crack diagram indicates that the
steel is yielding on average in the Y direction at this load level. Clicking with the mouse
on the control plot, or using the Page-Up and Page-Down keys allow changing of the
current load stage.
Also on the lefl, at the top, is a list-box that allows selection of which group of
nine plots to examine. By default, the "General" page shows up. Another page shows
Mohr's circles and a list of the fiil1 stress and strain state of the element.
To examine the data more closely fiom one of t h e plots, it is possible to right-
click on the plot and select "view data." This allows the data to be copied to another
application such as a spreadsheet to check the data or make other plots.
An analysis like this generally takes less than one tenth of a second. Tt becomes
possible to quickly find the effects of different reinforcing levels, for example, this way.
See the Appendix A for more information on Membrane-2000.
2-3Quick Start: Response-2000
Unlike the other programs, Response-2000 does not have a default cross section
entered into it. This is not a problem, however, as one can be made quickly. For this
example, a n 8 0 foot span prestressed concrete bridge girder and slab will be analysed.
First, as this example is presented with US customary units rather than the default
SI metric, select it from the "Options 1 Preferences" dialog box. To select US units as a
default each time the program begins, see section 5-1 1 o f Appendix A.
Secondly, g o to the "Define 1 Quick Define" dialog box. This is a "wizard" that
allows a section t o be created quite quickly, usually within 30 seconds. Each of the four
programs in this thesis has such a wizard t o make new files quickly.
The first page of the dialog box asks for a title and material properties. Mer
entering a title, say, "Test Section" with the reader's initials for the "Analysis by" box,
the material properties may be selected. For this example, the 5000 psi concrete, 60 ksi
steel and 270 ksi strands are fine, so select the 'Wext" button.
The second page of the wizard asks for the concrete cross section. At the top of
the list are simple sections such as rectangles and circles. In the rniddle of the list are
more exotic shapes such as columns with interlocking hoops, and hotlow columns. At
the bottom are the "standard shapes" such as AASHTO girders. As this is what is needed
here, scroll down near the bottom of the list and select "Standard Shapes AASHTO.
Press tab (or click with the mouse) to the right side to select the type of section. Pressing
any key will pop up a seleaion box to select a section fiom the currently defined listings.
Select the AASHTO Type IV girder and press "ok". For the next input field, enter zero,
as there will be no "haunch" on this section (i-e., no extra concrete between the top of the
precast beam and the bottom of the slab.) Select a slab depth of 8 inches, and a slab
width of 80 inches, and select 'Wext" to go to the next page of the wizard.
The third page allows selection of the longitudinal reinforcement for the section.
The top half defines bars in the slab for this standard cross section case and the bottom
defines non-prestressed steel in the bottom of the cross section. Leave the default of 20
#4 bars for the top, but remove the 3 #8 bars for the bottom by entering "0" for the
number of bars in the bottom half of the screen. Press the 'Wext7'button again to go to
the last page of the quick menu.
The Iast page allows selection of the stirmps as well as the strands. Select "open
stirrup" fkom the list of stirrup types. The default bar type of #4 is reasonable. Select a
spacing of 16 inches. Switch the clear cover to 2 inches fiom the default value of 1 . 5 7 ,
"
which is actually 40 mm converted to inches. Finaily, enter 30 for the number of strands.
The prestrain listed as 6.5 represents a jacking stress of 70% of ultimate, and is therefore
reasonable. Select the "Finish" button to complete the definition of the section.
I
-(Id) 64-1 maS.1
Y, ml 222 226
Y, FI 39.8 30.4
Crack Surcinp
2 x ~ + 0 . 1 ~ 1 p
The analysis should take about 10 seconds to reach the peak load, and then about
20 more seconds to daennine the post-peak ductility for the section. The following 9-
plot screen will show up. These plots represent the state of the beam at failure, as shown
by the location of the crosshairs on the control plots. Each plot is drawn with respect to
the depth of the section. For example, the top centre plot shows the longitudinal strain
versus depth for the section showing the basic assumption that plane sections remain
plane.
The cross section in the top left is drawn darker in regions where it is predicted
not to have cracked. In this case, only the web of the beam is predicted to be cracked at
the shown failure Ioad. The top right shows the variation in transverse strain over the
depth, with a maximum of 7.3 parts per thousand near the top of the web. The crack
diagram shows the predicted angle and width of cracks in inches. The shear stress plot
shows that the shear is not unifonnly distributed over the depth of the section, though it is
fairly constant in the web at about 630 psi.
The bottom lefl plot of the 9 plots shows the principaI compressive stress values.
The red line at the lefi of the plot is the maximum allowed stress versus depth and the
right blue line shows the applied stress. Note the shear has applied an additional diagonal
compression in the web on top of the expected concrete stress profile fiom the
prestressing force. The two lines on this plot are about to touch at the top of the web,
indicating that this section is about to fail by crushing at the top of the web.
The two control charts show that the "V-Gxy" curve, that is, the shear force-shear
strain plot, is descending with increasing shear strain, whereas the lower control chart, a
moment curvature plot, is unloading along its loading curve. This indicates that the
section is predicted to fail in shear. The maximum predicted shear capacity of the section
is 249 kips. By scaling this fiom the loading, it is predicted that the beam would fail in
shear at this location if the applied load were to increase to a level of 23 kipdfoot.
2-3-4 Member Response
Response-2000 will calculate the fll member behaviour for a prismatic section as
well. To get a prediction of the behaviour of this 80-foot beam, such an analysis will be
performed with the beam subjected to a uniformly distributcd load. Fust select the "Load
1 Full Member Properties" menu option. Select the "length subjecte to shear" at the top
as 480 inches. (The analysis is done fiom one end to the mid-span of the beam.) Also,
select in the top options a uniform distributed load rather than a constant shear analysis.
This is the second option in the top list of three buttons. Click "ok" and select the "Soive
1 Member Response" option.
When the analysis is complete, the screen will change to the deflection page as
shown below. The top diagram is the predicted crack pattern at failure for the entire 40
foot half-span o f the beam. The bearing support plate at the Ieft bottom can be seen and
the right side represents the midspan of the beam. Estimated crack widths are shown in
inches. In the top control plot at the lefi is the M-V interaction diagram, which also
shows the combinations o f applied moments and shears along the length of the beam as a
red line. For a u ~ f o r m l ydistributed load, such as this, the majority of the loading line is
a parabola, with the shear reduced to zero near the support due to non-sectional load
resistance mechanisms in this region. The shape of this load diagram is explained in
Chapter 7. It can be seen fiorn the interaction diagram that the loading envelope is
touching the strengh envelope almost simultaneously at the right side bottom (flexure in
positive moment at midspan), as well as at the top (shear near support). Indeed, the
midspan cracks are predicted to be almost 1 inch wide, and there is substantial shear
cracking (0.147 inch cracks) near the support.
The bottom control plot shows the predicted load-deflection relationship for the
beam (pushover analysis results for column analyses). The final behaviour is predicted to
be fairly ductile, with a 22.9 inch deflection at a failure load of 23.4 kipdfoot. Assuming
that the load capacity is acceptable, this would seem to be a fairly efficient design in
t e m s of shear versus flexural capacity; more stirrups would not be needed, as the beam
would fail in flexure first. A lower amount of stirmps would subject the beam to a
potentially brittle shear failure, however. In a design like this, it is wise to err on the
conservative side of shear design and include somewhat more shear reinforcement than
what has been provided. Of course Response-2000 allows any such option to be
conveniently checked by changing the spacing of the stirnrps, and rerunning the analysis.
2-4 Quick Start: Triax-2000
L 0 8 d h (dNx.dNx,dVw + w . d w . w
o.m. o.m. o.m. 1.m. am. am
sufficient size in al1 three dimensions to cover a senes of cracks.
Load Direction Nx N, N, V, V
, V
,
Load Ratio 1.0 2.0 3.0 4.0 5.0 6.0
This loading represents triaxial tension on the element as well as increasing shear
in al1 shear directions. These load ratios are entered into the program by selecting the
"Loads 1 Loads" menu option. As in each of the programs explained in this thesis, there
are two colurnns of numbers that may be entered. The left column is for the load level at
which to start the analysis and the right column is for the loading ratios to be used for
incrementing load after that point. Note that the actud vaIues on the right column do not
matter, only their reIative values and signs are used in the program. Enter the above load
levels into the right side column of the loads menu and close the loads dialog box by
clicking the "ok" button.
On clicking the "solve" button on the toolbar, the now familiar nine plots show up
with the results of the analysis as shown below. The control plot is autornatically
selected by the load ratios and in this case shows the load-factor vs. shear strain in the Y-
Z direction.
L a d FwtorCrwL wl
P V
Tnax-2000 shows a tabular list of al1 the strain and stress states for the element at
the load marked by the crosshairs on the control plot. The crack diagram shows the
principal directions as well as the intersection of the crack planes with the outside of the
concrete volume. In general, 3D behaviour of this type requires some study to ensure
that the results are indeed what is expected.
2-5 Quick Start: Shell-2000
The last of the four programs in this thesis is Shell-2000. It
assembles a collection of Triax-2000 elernents on top of each
other to allow out-of-plane analyses of plates and shells. As
such, it is a three dimensional analogue of Response-2000. It is
a more general version of Membrane-2000 that will allow
analyses that include out of plane forces. Such shell elernents
can be found in slabs and walls and, indeed, almost al1 structures made of plates or shells.
Loading for Shell-2000 consists of the following eight force resultants: Axial
force in X and Y directions, moment about X and Y axes, out-ofiplane shear about X-Z
and Y-Z planes, twisting moment (M,) and in-plane shear. Shell-2000 is a superset of
Membrane-2000 and can do al1 analyses that Membrane-2000 can do. Due to the
inherent 3D nature of the implementation, however, it is slower than Membrane-2000.
Ail diirmiora in d l S m t r i s
M i i h r n c k i r covw : Il mm
SE4
R e H o r c m a stress 2
b Compressive Stress
It c m be seen from the control plot that failure is predicted to be fairly ductile.
From the bottom line o f the output window, the failure in-plane shear is predicted to be
976 kN/m. In the test, the element failed in a ductile fashion at an in-plane shear o f 961
W / m . The nine plots show the state of the element at failure. The steel is predicted to be
yielding on the top and bottom of the shell in the Y direction a s well as in the bottom side
in the X direction. The crack plot shows that the element is predicted to have full-depth
cracking, roughly in the X direction at the top (flexural compression side), and rotated
through the depth as a resuit of the in-plane shear stress. From the principal compression
plot, the concrete is predicted to be crushing (two Lines touching) at the top due to the
flexure as well as at the bottom due to the in-plane shear.
Chapter 3: The Modified Compression Field Theory and
Related Numerical Techniques
3-1 General
Perhaps the most important differentiating element between different sectional
models is the constitutive models that are employed. The programs in this thesis use the
Modified Compression Field Theory (MCFT). This theory traces back through the
Compression Field Theory of 197814to the Diagonal Compression Field Theory of
197413.Vecchio defined the original fomi ofthc MCFT in 1982'' fiom the testing of 30
reinforced concrete panels subjected to uniform strain States in a specially built tester.
The definitive description of the MCFT is in the 1986 Arnerican Concrete Institute paper
"The Modified Compression Field Theory for Reinforcd Concrete Elements Subjected
to ~ h e a f " . Since then, only two small changes have been made, both in 198716.There
have been other proposed changes to the theory, but at the University of Toronto, it is
largely this 1987 version that is still used 13 years later. Since then, others have proposed
similar models including Hsu and han^"*^* and Kaufmann and ~ a r t i ' ~To. allow
cornparison between some of these proposais, the methods of Hsu et al. have been
expl icitly included in Membrane-2000.
The most important assumption in the model is that the cracked concrete in
reinforced concrete can be treated as a new matenal with empirically defined stress-strain
behaviour. This behaviour can differ fiom the traditional stress-strain curve of a cylinder,
for example. The strains used for these stress-main relationships are average strains, that
is, they lump together the cornbineci effects of local strains at cracks, strains between
cracks, bond-slip, and crack slip. The calculated stresses are also average stresses in that
they implicit1y include stresses between cracks, stresses at cracks, interfce shear on
cracks, and dowel action. For the use of these average stresses and strains to be a
reasonable assumption, the distances used in detemining the average behaviour must
include a few cracks. Sectional models satisfLthis by needing to be at least a couple of
section depths long.
Figure 3-1 summarises the Modified Compression Field Theory for the two
dimensional case. The lefi panel shows the equilibrium equations which are in fact
simply the equations of a Mohr's circle of stress. The middle panel shows the strain
conditions, which also can be summarised by a Mohr's circle. Note the irnplicit
assumption in the MCFT that the angle of principal concrete stress can be taken as equal
to the angle of principal strain (0). The final panel shows the stress-strain relationships
for reinforcement, concrete in compression, and concrete in tension. The bottom of each
panel shows the components of the crack check to ensure that the average stresses can be
transferred across the crack.
-
Figure 3 1 The Modiiied Compression Field Theory
was given as ,/m. The change of the coefficient fiom 200 to 500 was suggested by
Collins and Mitchell in 198716partly as a result of examining expenmental results from
larger elements than the 30 panels tested by Vecchio. Note that later in this thesis, the
above tension stiffening model will be extended to more directly account for bond
behaviour.
The three stresses in the 2D case (f,, f,, v,) extend to six (f,, f,, f,, v, v, v,).
The three strains in 2D (ex,+, y,) extend to six as well (E, +, G,y, y
,, y,).
The equilibrium equations are extended simply to include one additional equation
for the Z direction. Note that the 2D tensor format (Mohr's circle) still applies as a 3D
tensor. The angte theta in the 2D case extends to a set of 3 direction cosines for the 3D
case (k;,li, mi). As in 2D, the 3D direction cosines of strain are assumed to be the same
as the direction cosines of concrete stress.
Direction Cosines:
kl, 11, ml
k2, 12, m2
k3, 13. m3
Note 6-orn this that the concrete stresses are generally compressive in the x, y and
z directions when shear is applied as they will be balanced by the steel, itself in
tension. Again, note that the rotation is provided with the same direction cosines
used for the strain tensor. The directions of principal strain and concrete stress are
assumed to coincide.
I g) ifcalculated
desired, steel stresses at a crack and shear on crack components may be
as explained in Chapter 4.
a = E.-(E) xE
where the secant modulus, Lt
is a fnction of s."
These secant moduli are shown for concrete and steel in Fig. 3-2 and 3-3.
O O-5 1 1.5 2 2s 3
Principal Compressive Stnin (mmhn)
-
Figure 3 2: Secant and Tangent Modulus of Concrtte
-
Figure 3 3: Secant and Tangent Modulus of Reinforcement
Using this scheme, the strain vector {E) is related to the stress vector {a)via the rnatrix
ID]by the following:
Using this simple matnx equation, the solution for any unknown term can be
found simply and with stability. The secant stiffness matrix is symmetric and hlly
populated.
Pl = P c l + C P S I 1
To determine the CDJ matrix, it is first necessary to calculate it in principal
directions and then rotate it back to the X-Y-Z fiame. This rotation is performed with the
following relationship:
k, 1, and m are the direction cosines as noted above with the numencal example.
Note that the matrix can conceptually be divided into 4 quarters, each a 3x3 matnx. The
top leA corner deals directly with effects in the axial directions to strains in the axial
directions for example. The other quadrants include shear effects and so include at least
two terrns each as they will be affected by the two directions that make up that individual
shear strain.
The matrix CD,]' is the stiffness of the concrete in principal directions defined as
follows:
[Dc]' =
& and & are principal secant stifiess values as follows:
Where k1,
Because reinforcement is considered to oniy accept axial force, the matrices CD,]
for the X Y and Z directions al1 have only one element. D].for example has a term in
the top lefi corner equal to &.rhox where
L=fsxk (3 -8)
The Y and Z directions are similar. Note that these are not tangent stifhesses.
For 2D analyses, the [P,] matrix can also be written in terms of sines and cosines
of the angle of principal stresdstrain 8. For that case, use only columns and rows 1,2 and
4 in the above matrices, and make the following substitutions:
Clearly, if the concrete is in tension, the tension equation is used (which is only a
function of the principal tensile strain) and if the concrete is in compression, the
compression curve is used. Note that the compression curve is a hnction of two strains,
the principal compressive strain (e2) for the location on the curve (fi) and the principal
tensile strain (el) defining the height of the curve for the sofiening (f-). For each row
of the final tangent stifiess matrix, say the first row, there will be a unique relationship
between the change of siand the change of et. In 2D and for the first row of the final
matrix, any increase in ~2 will also increase by a factor of tan2& This happens becaux
the first row of the matrix is with respect to changes in the e, strain. Any change in E,
will cause a change in both of the principal strains. For the second row, the rate of
change of el 4 t h respect to EZ is LI tan20, and for the third row, the rate is -1.0. This
makes the derivatives more complex to calculate. If the principal compression equation
in Fig. 3-1 is broken up as follows, the stiffness c m be calculated:
fi = fi, Parabola
This means that the symmetry of the secant method above is broken. The tangent
stiffness matrix is not symmetnc. To calculate it, it is necessary to determine the
derivative of si with respect to E* for each row and calculate al1 the principal stifhess
values. Then calculate the entire stiffness matrix and discard al1 of it except the needed
row. Then repeat for the next row. For the 2D case, the matnx can be expanded
algebraically, reducing the amount of unnecessary calculations. For the 3D case, the
sirnplest implementation is to numerically calculate the derivative and go through the
matrix multiplication six times, once for each of the six rows.
Note that if the concrete principal tension is being affectai by the crack check as
explained in chapter 4, it is necessary to include the stifthess of this crack check in the
matrix as well. That is, the stiffiiess of the principal tension direction should be
controlled by the crack check equations, if they govern, rather than by the base curve
equation.
The shear stiffhess tenns in the i and j direction are calculated as follows:
Were it not for the increased speed and stability of the longitudinal stifiness
method over the traditional numerical methods, it would be dificult to justiQ this
complexity. In fact, the increase in stability results in the prograrns running 5-15 times
faster with the tangent stiffness matnx calculated this way over the older numerical
methods of calculating tangent stiffness. It is suggested that this does indeed justifi the
complexity of the prograrnming.
Chapter 4: The Crack Check
4-1 General
The crack-check in the Modified Compression Field Theory (MCFT)represent
an explicit check to ensure that the average stress levels can be resisted locally at a crack.
It has becorne apparent, in the past, that some researchers and engineers have
implemented the MCFT without including the crack-check. This is unwnservative and
potentially unsafe. In a series of papers, for example, Hsu has shown that ignoring the
crack-check can produce resutts that are very unconservative indeed26.27 . it is felt that
the crack check is sufficiently important to warrant its own chapter, though it could be
argued that it belongs in the previous chapter that introduced the Modified Compression
Field Theory.
The need for the crack check is easy to demonstrate. Consider the concrete pnsm
subjected to axial tension reinforced with 0.75% of 400 MPa steel as shown in Fig. 4-1.
The total force on the element may be calculated as follows:
N=Nc+Ns (4-1)
-
Figure 4 1: Prism in Tension
The stress-strain relations for average concrete and steet behaviour are defined by
the usual MCFT equations as shown earlier in Fig. 3-1, and again here in Fig. 4-2 and 4-3
for concrete and steel respectively
-
Figure 4 2: Average Tensile Stress fl -
Figure 4 3: Average Steel Stress f,,
A nave analysis may produce the graph in Figure 4-4 for the total stress versus
strain relationship. Note that the concrete and steel forces have been added together over
the entire range of strain. This is not correct.
Concrete -
O 1 2 3
Strain (mmlrn)
-
Figure 4 2: Total stress with no crack check
Consider the free body diagram shown in Figure 4-5 with the leA half of the
element drawn with average stresses as used in the MCFT and the right half with local
stresses at crack where there will be no concrete tension.
fsx
Average
fi
-
- fwcmk
At a crack
Recalling that the MCFT,for simplicity, uses the same stress-strain curve for steel
at a crack and on average, it is possible, indeed cornmon, for both f,, and &crack to
equal fv, the bare-bar yield stress. From the fiee body diagram, it is clear that the
concrete tensile stress, fi, must equal zero in this case. Ensuring that the local stresses at
a crack not exceed the yield stress in this case is the job of the crack check. For
reinforcement with, for example, a biaxial stress-strain response, it can be assumed that
the stress at a crack will always be able to achieve at least the stress corresponding to the
bare-bar stress at the given average strain.
Using this crack check results in the correct total stress plot prediction shown in
Figure 4-6 for this element. This corrected answer includes the effect of the crack check
to ensure that the steel stress never exceeds the yield stress of the bare bar at a crack
-
Figure 4 6: Total stress with crack-check
4-2 Crack Check in One Dimension
The above demonstration implicitly showed the crack check in one dimension.
The equation that must be satisfied is simply a reorganisation of the above relationships,
namely :
Consider the fiee body diagram, shown below in Fig. 4-7, of a two dimensional
piece of reinforced concrete drawn at a crack on the top right and on average at the
bottom left. Note that the cut is at an angle theta, the same angle as the cracks, principal
strains and concrete stresses in the MCFT. Note that the force arrows for the
reinforcement are only drawn once, rather than once per bar, to maintain simplicity.
-
Figure 4 7: Two dimensional crack check
This assumption of minimising shear on the crack has the effect of "using up" al1
the steel capacity in the weak direction before any shear on the crack is required. As this
behaviour is only happening locally at a crack, it will have no effect on the overall stress-
strain state unless it requires the lowering of the princpal tensile stress.
Sumrning forces in the X and Y direction on the above diagram results in a series
of equations that define the crack check. These equations have been summarised in
Table 4-1, which includes al1 the steps needed to saisQ the crack check in two
dimensions. Following these steps will ensure that the stress at a crack is not in excess of
yield in either the X or Y directions and that the shear on the crack is less than the limit
based on crack width. The average concrete tensile stress thus calculated (f~),can be
used with the rest of the MCFT equations as presmted in Chapter 2.
6, = fi( fv- fa) This is the extra average stress required in the
fiw = fi ( f ~~-S Y ) X or Y diredion ta yield the longitudinal steel.
2) Biaxial yield without shear on crack (ffb)
flb =fia ms20+ fisr sin2e Ensures that load required to cause biaxial yield
of reinforcement at a crack is not exceeded.
3) Max shear on crack fi,f biaxial yield ( v d
vd = 1-,lf ,if 1 sine cos0 Calculate the shear on the crack required to
Achieve #axial yield of the reinforcement
$) Calculate maximum allowed tensile stress Rom X and Y dhction equilibn'um
frc= f,i + min( , v d ) cote Calculate maxm
i um tension from equilibrium
fld = fw
l + min( ver , v d ) tan0 fi, is for X dir'n, fid for Y dir'n equilbrium
5) Take minimum of al1 calculated values for final value of pnhcipal tension
fi = min( f~,, f i b , f i e , fid) Sefed Minimum Value
To calculate the actual equilibrium-based steel stresses at a crack and the value of
the shear on the crack (vd), use the following steps, using the notation in Table 4-1. This
cannot be made into a simple equation because, as noted above, there are more unknowns
than equations. By using these steps, the shear on the crack (vk) will be defined,
allowing the steel stress at a crack to be calculated fiom the equations in Fig. 3- 1.
Unlike the 2D case, calculating the shear on the crack and steel stresses at a crack
for any given load level is cornplex. Recall that in two dimensions, there were 2
equations and 3 unknowns required to calculate the behaviour at a crack. In three
dimensions, there are 3 equations and 5 unknowns. The equations are the sum of forces
in the X, Y and Z directions. The 5 unknowns are the reinforcement stresses at a crack in
the X Y and Z directions along with two components of the shear on the crack. Shear on
the crack in three dimensions is a 3D vector, but it is known that the shear on the crack
must be in the plane o f the crack, removing an unknown.
A simple algonthmic method as used above will not sufice for this case. An
elegant way to solve the problem is to redise that the calculation of the answer is in fact a
nonlinear optimisation problem of quadratic programming. Once the shear on the crack
components have been found, the steel stresses may be calculated simply with equations,
similar to the 2D case above. While there are only two components to the shear on the
crack as the vector must be in the plane o f the crack, it is simpler to Ieave the solution in
terms of three components. The requirement that the vector remain in the plane of the
crack will be maintained with the form of the equations used. As such, the problem is to
find the 3 components of the shear on the crack subject to the constraint that the steel not
exceed yield in any direction and that the shear on the crack be a minimum.
Using the assumption that the goal is to minimise the shear on the crack, and the
notation in Table 4-2, the following quadratic programming problem arises:
2 2 2
Minimise vci2 = V ~ W + V ~ Y Z +- V Q ~ L
Each of these constraint equations is one of the biaxid yield equations fiom Table
4-2. With only three degrees of fieedom and nine constraints, this problem can easily be
implemented using the brute force approach of checking each constraint individually and
then by checking the intersection o f different constraint planes. Note that the traditional
simplex method of linear programming may not be used, as this is a nonlinear problem.
Once the shear on the crack components are calculated, the steel stresses at a
crack can be calculated with the following equations using the notation in Table 4-2:
Average Average
Cross Section Longitudinal Strain Concrete Stress Steel Stress
(mmfm) (MW (MW
-
Figure 4 8: Btam in Fltxurt
Figure 4-8 shows a rectangular beam in simple flexure. Assuming that there is no
direct concrete tensile stress transferred across a crack, moment equilibrium about the
neutral axis produces the following equation:
That is, the moment about the neutral axis caused by the average cuncrete tension
must equal the moment caused by the difference between the steel force on average and
the steel force at a crack.
The lefi side of the equation above is calculated in two steps. Figure 4-9
describes the terms used in the calculation. The goal is to determine the largest average
concrete stress component that can be tolerated without yielding the longitudinal
reinforcement. A moment (Mi) is first calculated assuming a concrete stress of twice the
cracking stress (ft) at the neutral axis and zero at the bottom of the cross section. This
variation in concrete tensile stress is integrated by taking moments about the neutral a i s .
A second Moment (M2)is then detennined as the remainder of the capacity moment afler
MI is subtraaed. A maximum value of average concrete tensile stress at the bottom of
the cross section is calculated that would cause this moment. If the wncrete exactly
achieved the stress profile fiom 2 fr at the neutral axis and fi, at the bottom of the beam,
the longitudinal steel would just yield at a crack.
'1 max
Cross Section Average Conuete Tensile Stress
-
Figure 4 9: Conerete Moment
The reason for using twice the concrete cracking stress as an allowable stress near
the neutral axis is that for cross sections without transverse reinforcement, the
longitudinal tensile demand immediately below the flexural cracking fiont can exceed the
cracking stress. That is, the horizontal component of the principal tensile stress added to
the horizontal component of any shear on the crack sums to more than the cracking stress.
Though this demand is high, the reinforcing steel below the crack fiont will, in general,
have sufficient capacity to equilibrate this. If it does not, then this technique will lower
the allowable tension to ensure it satisfies equilibrium. The use of two components to the
flexural crack check is necessary as a single component was found to be numerically less
stable. The two components more smoothly transition fiom a case with plenty of extra
capacity at a crack and no capacity at a crack.
As this process does not require the results of biaxial or triaxial stresses, the
flexural yield crack-check performed in this way is non-iterative.
There are some restrictions implicit in this method.
1) The concrete compression block is totally ignored in the analysis.
2) The assumption that a bar is always able to have a main concentration factor of 1.25
or 2.0 at a crack means that bars that are linearly elastic up to failure are predicted to
rupture at a strain well below the real rupture strain and stress. Further research is
needed to develop more appropriate methods for such materials.
3) If there are no ban crossing the cracked zone, equilibrium requires that there be no
average concrete stress. This was found to badly afEect convergence at first cracking,
however. As such, if there are no bars crossing the crack face, the concrete stress is
not reduced. This is done for computational stability despite flagrantly violating
equil ibrium. The programs in this thesis are, therefore, not appropnate for calculating
the flexural behaviour of plain concrete specimens. A case where this is more
common is in the response of a singly reinforced section subjected to negative
bending.
4) It is possible that in attempting to yield the top bar in a flexurally cracked zone, the
bottom bar will rupture before the top can yield. This means that the calculated
capacity concrete tension moment cannot be achieved. This is not dealt with
explicitly, but shows up in an inability to caiculate stresses at a crack. This generally
has only a small effect on practical cross sections.
Due to the possibility that some of the bars may become nonlinear during this
calculation, it must be iterative. To Save time, the stress at a crack is only calculated once
per load stage, just before showing it to the user.
Chapter 5: Concrete Constitutive Relations
The Modified Compression Field Theory (MCFT)treats cracked concrete as a
new material, so it is necessary to define the stress-strain characteristics of this new
material. As the MCFT is based on principal stress-principal strain relationships with a
special check for shear on the crack the relations may be divided into the categories of
be haviour in compression, behaviour in tension and behaviour in interfacial shear.
There are a series of base cuwes available in the programs for modelling concrete.
These are explained in Section 2 of the program's User Manual in Appendix A. The
recommended base curve, the Popovics cuxve, is shown in Figure 5-1 for a series of
different strength concrete cylinders. This base cunre was defined by ~ o p o v i c s 'and
~
partial1y calibrated by ~ o r a s z is
~ 'shown in the following equations (in MPa units):
-
Figure 5 1: Popovics Concrete Base Cuwe -
Figure S 2: Parabolic Concrete Base Curve
For both of these curves, the strain at peak stress (ES) is needed to complete the
definition. This is a function of the shape of the base curve as well as the initial tangent
stiffness of the concrete &),itself a finction of the stiffness of the aggregate. Ifthe
initial tangent stifiess of the concrete is known, or a stress-strain curve from a cylinder
test is available, an eaimate ofthe arain at peak stress may be made. Ifneither are
available, then the following method is suggestedS(in MPa units)
Figure 5-3 shows this relationship compared to a selection of data representing the
peak stress and grain at peak stress for test cylinders associated with shear tests fiom the
University of ~ o r o n t o28' ~and the University of oust on'^* 'O* ". The data frorn the
Toronto tests give higher than predicted ec' values as Toronto's crushed limestone
aggregate tends to produce lower stiffness concrete than that predicted by the equation
above.
O Houston qiinden
+ Toronto cylinders
-
Figure 5 3: Strain at peak stress
-
Figure 5 5: Compression softening: High stnngth concrete
59
5-2 Bebaviour in Tension: Uncracked Concrete
Concrete in tension is assumed to act Iinearly until first cracking wit h a stifhess
equal to the initial compression tangent stifhess. Cracking is clearly an important
phenornenon in concrete. Traditionally, the use of the AC1 shear cracking stress of
4 E (psi), 0 . 3 3 z (MPa) has been suggested for use in the MCFT. However, it has
been found that this prediction of cracking strength is not particularly good for high
strength concrete. Figure 5-6 shows the cracking strength versus concrete compressive
strength for a senes of large reinforced concrete elementsL1*
"* 'O* "* "* 33 dong with the
AC1 relationship.
A C 1 Shear Crackhg
/ ( O Houston Shell
Average = 0.90
C.0.V = 24.6%
O ! 1
O 20 40 60 80 100 120
Concrete Compressive SWenglh (MPa)
-
Figure 5 6: AC1 Shear cracking strength o f concrete
The fit to the higher strength data is not especially good. Instead, the programs in
this thesis al1 use the following relationship for the cracking strength of a large volume o f
concrete (MPa units):
ft = 0.45(f,')O.~ (5-3 )
Figure 5-7 compares this equation to the same data set used above.
f, =0.45-(fc )0.4 A A 00 0
O Houston Shell
ExpPred: n = 83
-rage = 0.98 A Toronto Shell
C.0-V. = 23.5 % + Toronto Panel
O ! I I 1
O 20 40 60 80 1O0 120
Concrets Cornpresshm Sbength (MPa)
-
Figure 5 7: Proposed shear cracking strength o f concrete
Figure 5-8 compares the proposed cracking strength equation t o the traditional
AC1 equation. It can be seen that for concrete strengths less than, Say, 40 MPa, there is
only a small change in the predicted cracking strength. For 120 MPa concrete, however,
the predicted cracking strength drops by 18% below the AC1 value. This is imponant in
predicting the behaviour of high strength concrete beams without stimps.
-
Figure 5 8: Cornparison of shear cracking strengths
Note that the relation proposed is appropriate for structural analysis o f a large
volume of concrete. The data points used to derive it are fiom uniformly loaded elements
containing up to 1 m3 of concrete volume. Tests involving small volumes of concrete o r
with high strain gradients through the depth can be expected to show higher strengths.
For example, the modulus of rupture test on a 150 x 150 mm cross section only exposes
one one-thousandth as large a volume of cancrete to high tensile stresses as the large
element tests do. The chance of the weakest, and hence controlling, piece of concrete
being in this small volume is remote, meaning that strengths fkom the modulus of rupture
tests are generally higher than the above equation would predict. It is recommended that
even if such results as modulus o f rupture o r spiit cylinder strengths are available for a
given concrete, the above equation be used.
The first equation is that proposed by ~ecchio". based on his original 30 panel
elements.
The second equation was proposed by Collins and Mitchell16, based partly on the
results of the large shell element tests.
The third equation presented here is the ~ a r n a i ~equation
' as used by Hsu and his
colleagues at the University of Houston.
Strain ( M m )
Note that there is a substantial variation between these predictions. Generally, the
1982 equation fits well to tests in the small University of Toronto panel tester, the 1987
relation fits well to data from the larger University of Toronto shell element tester, and
the Tarnai relation has done a reasonable job with the data from the University of
Houston. If an equation of the same form of the MCFT is fitted to the Houston data
originally used to justify the use of the Tamai equation, a coefficient of 1500 is produced,
rather than the 200 and 500 in the other equations. It has been something of a mystery
why these differ as much as they do, though various explanations such as overall size
have been used.
Consider that tension stiffening is largely a bond phenornenon. That is, it is the
bond between the reinforcing bar and the concrete that causes any tension to develop in
the concrete between the cracks. Collins and ~itchell', for example, reommended that
tension stiffening be reduced by a factor of 0.7 for plain bars or strands which will have
poorer bond properties than defomed bars. It is proposed that concrete tension stiffening
should be made a more specific fiinction of bond characteristics of the reinforcement.
Thus, at locations where the concrete is reinforced with a closely spaced array of small
diameter bars, the average tensile stress in the cracked concrete can be expected to be
higher than at locations reinforced with a widely spaced array of large diameter bars. The
typical cross-sections tested in the three research program mentioned above are shown in
Fig 5-10. Note that the bond characteristics of each will be different.
-
Figure 5 10: Scale of element tests
For the typical elements shown in Fig. 5-10, this bond parameter works out to be
62 mm for the University of Toronto panel tests, 163 mm for the University of Toronto
shell element tests, and 42 1 mm for the University of Houston shell element tests.
If the coefficient of the MCFT style tension stiffening equation is plotted with
respect to this bond term, the plot shown in Fig. 5- 11 is prduced.
Coefficient = 3.6 m 0
0 #=@
0@
0 HoustonShell
0#
Elements
0
##
.-
&e
0&
0 I
TorontoPanel
Elemento _, Toronto S M
flC Elements
0 250 0 we
-
Figure 5 11: Tension stiffening vs. m paramctcr
There is a clear relationship suggesting that for elements with poorer bond
properties, i.e. larger m values, the tension stiffening should be lower.
The above relation is based on tension stiffening equations that were averaged
from a number of elements in each series. By plotting the trend with individual
Based on this, it is proposed that for uniformly reinforced concrete, the coefficient
in the denominator o f the MCFT tension stiflening equation normally shown as 500 be
replaced by the term 3.6m as follows:
C
This will improve the scatter of the predictions of uniform elements subjected to
shear, as shown in Chapter 8. Note that this equation will only affect the behaviour
between first cracking and when the crack-check begins to control the behaviour. The
fact that this parameter has units indicates that it will be dependent on the absolute size of
the specimen tested. It is suggested that this term p a d y explains the size eflect in shear
as shown in Chapter 10.
For biaxially or triaxially loaded elements, there will often be different values of
the m parameter in the different reinforcement directions. It is recommended for these
cases that the selected value of the m parameter be the lowest value for each of the
orthogonal reinforcement directions. This means that the tension stiffening will be
controlled by the direction that has the best bond properties. This was found to
reasonably mode1 the experimental behaviour.
5-4 Behaviour in Tension: Poorly Reinforced Cracked Concrete
The equation proposed above is appropriate for regions of reinforced concrete that
are relatively close to reinforcement, such as in the membrane element experiments
describeci. Further away fiom reinforcing steel, however, the proposed equation is not
appropnate to use. Figure 5-13, bpxd on Collins & h4itchel15, shows the CEB-FIP
suggestions for the effmive concrete embedment zone. Concrete in this zone, within 7.5
db of longitudinal steel, can be assumed to have a significant average tensile stress
between the cracks. Outside this region, it is assumed that there is not any significant
average concrete tende stress afler cracking. That is, it would be appropriate to use the
tension stiffening relation proposed above in the embedment zone, but not outside of this
region. This simplification is appropriate for analyses without shear, but less so when
shear is considered. In beams without transverse reinforcement, for example, it is the
average concrete tensile stress that must balance the diagonal compression in the web.
Selecting a tension stiffening equd to zero would then imply that there is no shear
carrying capacity in a beam where the web extends more than 7.5 bar diameters fiom the
reinforcement. Taylor, arnongst others, demonstrated that the webs of such beams can
cany in the order of half of the applied total shear force by aggregate interlock"- As
such, a method is needed to calculate an appropriate level of tension stiffening for
regions that are a fair distance away fiom reinforcement.
-
Figure 5 11: Tensile stress in c r a c k d concrete
To mode1 such conditions, consider the "tooth" of concrete between cracks shown
in Fig 5-14. It is reinforced with a single bar and loaded in axial tension via the bar.
Tension stiffening in this example will be an average concrete tensile stress in the
direction of the bar above and below the bar. A linear elastic finite element analysis was
performed for such an element assuming that the concrete mess at the elevation of the
bar was constant and just below the cracking strength.
-
Figure 5 14: Concrete between cracks
Figure 5- 15 plots the average concrete tensile stress versus the x-axis in Fig 5-14.
-
Figure 5 15: Tension stimning vs distance from bar
The stress values have been normalised with respect to the maximum values found just
above the bar. Also s h o w is a simple curve fit equation.
The analysis predicts that the average tension in the concrete will reduce to zero
as the distance away fiom the bar reaches about half the crack spacing. This is expected
from St. Venant's principal as the "stress bulb" will be symmetncal about each half of the
tooth's width. Beyond half the crack spacing away fiom the bar, the predicted average
concrete stress is in f a a slightly compressive.
To spot check this analysis, consider the typical concrete elements drawn in Fig.
5-10. The geometry of these is such that no point in the concrete is frther fiom a
reinforcing bar than about 0.25 to 0.3 times the predicted crack spacing. Based on the
average value of tensile stress in Fig. 5-15 over that range, it might be expected that the
total tension stiffening would be about 0.6 times the maximum value. The base MCFT
tension stiffening equation predicts that level of tension stiffening for a strain of about 1
mm/m which is about half of the yield strain. This suggests that the calculated values
fiom the finite element analysis are reasonable.
To attempt to directly use the results fiom Fig 5-15 in sectional analysis, a few
dificulties arise. The biggest is that the reaI crack spacing over the depth of a beam is
discreteIy changing. The predicted crack spacing, explained in Chapter 7, smoothly
increases approximately as 2 times the distance fiom a bar plus a value at the bar. As
such, as the distance away fiom the bar increases, the crack spacing a h increases. To
deal with such issues, some empirical curve fitting was necessary to make the above
finite element analysis directly connect with the programs. This fitting was performed
with Response-2000 on a series of 83 elements without shear reinforcement.
Figure 5-16 shows a portion of the bottom of a concrete bearn. The tension
stiffening (fi) at depth z in the figure is calculated as follows.
Where
Note that the curve fit equation in Figure 5-15 only comes into effect if the
location of interest in the concrete is a suficient distance fiom the nearest steel. This
distance is assumed to equal twice the crack spacing at the bar for simplicity. Nearer to
the bar than this, the equation is similar to the equation presented above for the average
value over a membrane element.
Note equation 5-9 represents the incremental value of tension stiflening. The
earlier equation in Fig. 5-1 1 models the average tension stiffening for al1 of the area of
concrete in Fig. 5-16, whereas this new equation is the value to apply at the location z
alone. To simply mode1 this difference fiom the average and the incremental value, a
factor of 2.0 is applied to the equation. It was found that the results are not sensitive to
this factor.
Table 5-1 shows the use of the proposed tension stinening equations for a given
strain state for a bearn. The beam is L metre deep, with 4-30Mbars 100 m m up fiom the
bottom face of the beam. For the given strain nate, the tension stiffening parameters are
calculated sshowing the distribution of average concrete tension over the depth of the
element. Chapter 10 compares the effect of these proposed tension stiffening equations
o n the quality of predictions of Response-2000.
Notes: As Iisted in Chapter 7, Crack spacing cannot exceed member depth for beams in
flexure
Ac terni cannot be l e s than the width multiplied by twice the bar diameter
The controlling type of tension stiffening is shaded.
.-
O
I
=U
0.75 Cracking -25 days
Y
-
Figure 5 17: Cracking strength vs. drying time
Figure 5-18 shows the same pattern for tension stiffening. In this case, the base
MCFT tension stisening relationship was compareci to the observeci tension stiffening. A
factor was calculated for the reported data points that indicated the relative value of the
obsewed tension stiffening with respect to the expected value. For example, if a test
showed results identically the same as what the base equation predicts, then it would have
a value of 1.O, while another test having an average tension stiffening twice what was
expected would have a value of 2.0. Plotting these factors versus the days between the
completion of curing and testing produces the following trend:
Days of drying
The new method presented here is an extension of the methods used by earlier
nonlinear sectional analysis prograrns37.6.53 extended to increase performance and
computational stability. The method is used in Reponse-2000 and Shell-2000 to calculate
a new estimate of the shear stress profile for a given load level. The prograrns assume an
initial profile and then use this procedure to calculate a new profile and iterate until the
assumed and calculated profiles are in agreement.
Section A Sedion B
-
Figure 6 1 :Shear stress calculation
This section of beam is dx units long, and subjected to constant shear V and no axial
load. The moment at section A is taken as M, and, due ta the shear, the moment at
section B will be higher, M + d M = M + V-dx. The assumeci linear longitudinal strain
gradient corn the moment will cause a longitudinal stress profile with compression on the
top and tension on the bottom of the cross section. Consider the shaded m i o n at the top
right of Fig 6-1 as a fiez body diagram o f the top of the beam, from elevation z up to the
top of the beam. It is subjected to a force on the left from the moment, but a higher force
on the right from the stightly higher moment. This requires a balancing force on the cut
plane of the beam, shown as H. Due to the summation of moments about a point
equalling zero, the shear stress in a horizontal plane at a point must equal the venical
shear stress. As such, the force H divided by the beam width and dx results in the vertical
shear stress on the beam at depth z. This is the same denvation used to produce
Jourawski's familiar relationship:
An implicit assumption in this theory is that plane sections remain plane it was
used to calculate the longitudinal stresses. Though the shear strains associated with the
calculated shear stress will warp the section, violating plane sections, the warping does
not affect the longitudinal stress gradient for regions of constant shearn.
Compression
Stress Profile
Stress Profile
at Section B
Tension
Cross Section Longitudinal Stress Shear Stress
Profile Profile
-
Figure 6 2: Intemals of shcar stress calculation
It is useful to examine in more detail what is happening in the description above.
Figure 6-2 shows the simple case of a linear elastic rectangular cross section subjected to
shear. The stress profile that will occur at Section A and B, as in Fig 6-1, are dniwn
together on the same axis for cornparison. As section A has a srnaller moment, it will
have a smaller stress profile than at d o n B where the moment is larger. The shaded
region represents the difference in stress profiles between the two sections. It is this
difference that defines the shear stress profile. Also shown is the shear stress distribution,
which for this case is a parabola. Note that the slope ofthe shear stress plot with respect
to depth is zero at mid-depth, and maximum at the top and bottom of the section. These
dopes are directly proportional to the diflerence in the longitudinal stresses at sections A
and B. That is, the shaded area on the middle plot at any given depth is proportional to
the derivative of the shear stress plot with respect to the beam depth.
While program SMAL is relatively stable, there remain situations where the
program will stop in the middle of operation and no longer converge. With small
adjustments in the input conditions, it c m often be made to converge. It was thought that
the use of the dynamic layering would reduce the incidences of this instability. On the
contrary, it was found that the Response-2000 becarne less stable than SMAL when
dynamic layering was implernented. Over a period of about a year, a series of issues
were discovered that indicated that the numerical methods used by SMAL were
inherently unstable in certain circumstances. These problems o d y became visible with
the precision that comes out of a dynamic layenng routine. For example, issues relating
to the exact depth of cracking will only be important in SMAL if the manually assigned
layer divisions happen to coincide with the calculated crack fiont. In cases where this did
happen, however, it is likely that SMAL would be unable to converge to a solution.
Three major problems were discovercd with the previous statesf-the-art methods
for shear stress profile evaluation.
They will be demonstrated with the 1
metre deep Tee-Beamshown. It has
1.2% of 400 MPa longitudinal steel
and 0.8 MPa of stimips. The
concrete strength was selected as 100
MPa to highlight the effects. The
applied loads are a moment of 1.5
kNm for every 1 kN of shear force.
SMAL96 predicts the shear capacity -
Figure 6 3: Samplc section
as 456 kN for this section with the
analysis taking 24 seconds. Response-2000
predicts a shear capacity of 480 W,
controlled by longitudinal yield with a
solution time of 5 seconds. The predicted
shear stress profile fiom Response-2000 at a
shear force of 456 ISJ, is shown to the nght
as well.
-
Figure 6 4: Shcar stress profile
The first problem encountered resulted fiom the necessity to numercally solve for
a desired value of axial load, moment and shear at both sections A and B in Fig. 6-1.
Numerical convergence always includes some error, but the shear, and axial load, must
be essentially identical between section A and B. Figure 6-3 shows a shear stress profile
generated with an earlier version of Response-2000 that used the older shear stress profile
methods as used by SMAL at a shear of 456 W. It can be seen that the shear stress
profile does not "close" at the top, that is, the shear stress is not calculated as zero at the
top and bottom face of the beam as required. This was caused by a difference in axial
load between sections A and B of only 2.2 W. This corresponds to an average stress on
the gross concrete area of 0.0065 MPa. The error in longitudinal strain that this axial
load represents is O. 1 micro strain. This means that methods like that in SMAL must
either solve to axial strains to a precision of 1 part in 10,000, or "smear" the error if the
shear profile does not close.
-
Figure 6 5: Error in axial load convergence
A second problem is that the depth of cracking for section A will be different
fiom the depth of cracking at Section B, due to the different moments at each section.
The effect of this is that a large jump in the shear stress profile is predicted in the region
between the cracking depths. Figure 6-4 shows the e f f ' for the example beam with a
shear of 180 kN. This spike does not appear to have any real physical significance.
More importantly, it is strongly dependent on the distance dx between sections A and B.
For some cross sections, this spike can make the predicted shear stress negative in the
middle of the web for positive shear, for example, which also suggests that it is a non-
physical artefact of the calculation.
1.4
-400 -E
Jump in Shear stress due to different
-500 -E depths of cracking
-
Figure 6 6: Error in crack depth
A third issue is that the distance between the sections A and B in Fig 6- 1 tums out
to be important in predicting the shear stress profile. SMAL suggests a value of d/6 for
this parameter, Response-2000 had used a value of about 10 mm for al1 beams, and now
effectively uses zero mm as a result of the new methodology in this chapter. Figure 6-7
compares the shear stress profile predicted by SMAL, the old version of Response-2000
and the new version of Response-2000 for a shear of 456 kN. Clearly this parameter has
a large effect on the predictions. SMAL predicts the largest shear stress near the bottom
of the section whereas the new method predicts it near the top of the web. The
predictions of Response-2000 tend to have smaller scatta than that of SMAL, suggesting
that the Response-2000 profiles are better than the S M A L profiles.
-
Figure 6 7: Effect of distance dx
Each of these three issues resulted in analyses with old Response-2000 taking
longer than they should have as a result of extra iteration, and, in some cases, preventing
a solution completely. While the increasing performance of cornputers has made the
eficiency issue less imponant, it does not help for cases when a solution is not possible
at al1 due to numerical instability.
The use of derivatives mean that the solution for only one location need be
obtained rather than the two needed for the earlier method. As there is only one section,
the axial forces and shears are guaranteed to match on each "side" of the analysis. A h ,
there is only one depth of cracking to consider. Irnplementing this method has allowed
Response-2000 to run 5-15 tirnes faster than it did using the older dual section analysis
style techniques, largely fiom a reduction in amount of iteration. Much of this iteration
was, in fact, partially induced by the dynamic layering routines that made it more stable
in solving to individual load levels. Program SMAL,which does not use the dynamic
layering is comparable in wall-dock speed to the new Response-2000 when it is able to
obtain a solution. As a result of the dynamic layering, however, Response-2000
calculates much more data for each load level. A typical SMAL analysis may have 20
layers in it, whereas a Response-2000 analysis may have 50 layers. The method
presented here scales directly to 3D as implemented in Shell-2000.
The load-deformation state throughout the depth of the element will consist of a
series of biaxial MCFT nodes. The first step in calculating the shear stress distribution is
to calculate the tangent stifhess in the X-Y-Gamma directions for each node as explained
in Chapter 3. Each node will result in a 3x3 matrix of stiffness, Ki:
Recall the basic assumption with sectional analysis methods that there is to be no
total stress in the transverse direction. As such, it is necessary to modiw the stifiess
matrix to account for this before using it in the longitudinal stiffness method.
Specifically, since:
and dNyrnust equal zero, simple algebraic rearrangement shows that the modified nodal
tangent stiflness matrix can be represented as:
b-d bof1
Ca -- C-- r i kl
- 1 IL--
h.d . h-f
1--
I-Lm n]
So that
This nodal stiffness now models the stiffness of longitudinal stress and shear
stress to longitudinal strain and shear strains while maintaining no change in stress in the
transverse direction. Note that as the basic tangent stiffness matrix (Ki) is non-
symmetric, the reduced stiflness matrix (Ki') will also be non-symmetric.
These local nodal stiffness matrices may now be integrated to produce the global
sectional forces tangent stifiess matrix J:
So that:
Where d~~ change in longitudinal strain at the geometric centroid of the gross
concrete area
d4 change in curvature
dy* change in average shear strain for section
dN change in global axial force
dM change in global moment
dV change in global shear force
These nodal stifiess terms are integrated over al1 the layers in the depth of the
beam. In the prograrns, this integration is done quadratically as the dynamic layering
routines automatically check for accuracy using that assumption. Using Ct+ notation
where J[0][0] is the top left matrix element, and the notation in Fig. 6-6, the summation
over dl layers may be done as follows:
L 1
Cross Section
-
Figure 6 8: Parameters for global stiffness matrix calcu1ation
Global Axial Force Stifiness
J[O][O] += (il-bl+ 4-j2-b2+ j3-b3) dz/6
J[O][l] += Qi-bl*zl+4-j2-b.z2+ j3-b3-z3)-dd6 z included for curvature
J[O][2] += (kl-bl-si+ 4-kt-b-s2+ k3-b3-s3)dz/6 s included for shear strain
Where: bl,b2,b3are the top middle and bottom widths of the section for the layer
Z ~ , Z Z , Zare
~ the depths in the section corresponding to the widths bl,b2,b3
dz is total layer depth = z3-zl
sl,sz,s3 are the multipliers of the average shear strain for the given depth.
Note that the shear strain is defined as an average value (the global
shear strain) as well as with a shape profile that varies over the
depth with an average of 1.O. These terms define the shape of the
shear strain profile.
j,k,l,m are the stifhess terms fiom K' as defined above for nodes 1,2,3
This global tangent stiffbess matrix calculated fkom the biaxial tangent stifiess of
the nodes serves two purposes. Firstly it can be used to solve for the next load stage for
an analysis. That is, it can be used to solve for the next global strain state estimate to use
to minimise the error in the force state. Secondly, and more importantly for this chapter,
it can be used to calculate the shear stress distribution.
Using the global stiffness matrix, the following matrix calculation is performed to
solve for a tangent "virtuai strain".
If this virtual strain is added to the in-situ strain in the beam, a new force state is
predicted that would have the same axial load, the same shear, but a moment that differs
by exaaly V.1 metre. This virtual strain profile, when multiplied by the longitudinal
stifiess tems over the depth, will produce the equivalent of the shaded area on Fig. 6-2.
This then directly leads to the shear stress as before. By selecting a moment increment of
V- 1 metre, it is as though the distance between sections A and B has been selected as 1
metre. In fact, due to the use of derivatives for the stifiess, it is effectively zero metres.
The next step is to calculate through the depth of the beam the equivalent of the
shaded area in Fig. 6-2. Whereas previously it was necessary t o calculate the difference
of the longitudinal stress at section A and section B of the lret body in Fig. 6- 1, this value
is now directly calculated as the virtual main times the appropriate stitniess. The value
calculated by this multiplication is the rate of change of shear flow with respect to depth
in the beam. At m y given depth, the rate of change of shear fiow with respect to depth
would be calcutated as:
Where Aq is the dope of the shear flow diagram with respect to depth
j and k are the top two terms fiom the K' matrix at depth z
z is the depth at this location
dsKd4, and dy,are the global Mmial strains fiom above
This change in shear flow is then integrated over the depth of the section and
divided by the local element widths to find the shear stress profile. Note that the tangent
stifiess of the longitudinal steel must also be included using the same virtual strain
concept with the tangent stiffiiess of the steel.
When implemented properly, the only way that this method can fail to find a
solution is if the determinant of the J matrix is zero. Ifit is zero because there is no
stiffness against moment, then no solution can be found. It has been found that this is
very rare, but can happen in the case of beams subjected to high shear with tull-depth
cracking and al1 longitudinal and transverse steel yielding.
Chapter 7: How the programs work
The previous chapters have provided background into the analytical methods used
in the programs. This chapter provides details of how the programs work internally so
that others may implement similar programs.
Both programs are based exclusively on the secant stifiess method in two and
three dimensions as explained in Chapter 3. For full load-defonnation analyses, a
variable is automatically selected as being the most critical, usually a shear strain, and
then that value is incremented in small steps. The other strains are then iteratively
determined with the secant stiffness matrix until the load ratios match the desired values.
It has been found that the secant stiffness method is surprisingly stable and can solve for
most solutions quite efficiently. For single load solutions, the same technique is used, but
the load vector is fully defined with the full strain state then determined iteratively.
7-2Response-2000/Shell-2000
Most of this discussion is written directly towards Response-2000. ShelI-2000 is
directly analogous in intemal structure. Response-2000 is based on iseries of biaxial
nodes integrated along a line through the cross section. The global strain state is made up
of the longitudinal strain at the centroid of the gross concrete cross % d o n (&,O), the
curvature ($), and the average shear strain (yw). As the shear stress profile, and hence
shear strain, varies over the depth of the cross section, a numencal profile is used that
modifies the average strain (y-) to produce the desired shape of shear strain through the
depth. This profile has an average value of 1.O and is zero at the top and bottom of the
cross sections. For uncracked concrete on rectangular sections, the profile is a parabola
with a maximum value of 1.5.
To solve for any arbitrary load stage or an interaction diagram requires a fair
amount of nested iteration. Figure 7-1 shows the general steps in the iteration procedures
used in Response-2000.
get-forces
Calculates M.N,V using
BINARY TREE
Calculate-sheargrofile
solves for new shear strain profile
-SECANT
-
Figure 7 1: Response-ZOO0 loops
Response-2000 has 4 major nested lwps. Note that the one fiirthest in is
effectively the same as Membrane-2000. This inner loop calculates the transverse strain
needed to ensure that there is no overall transverse stress on the nodes consistent with the
basic assumption of the sectional model.
The loop outside that, the get-forces loop, calculates the sectional forces (N, M,
V) on a cross section for a given global strain state. This is iterative as the cross section
is automatically divided up ("dynamic layering") in a binaxy tree fashion sirnilar to
Reference 38 to ensure that the stress levels are interpolated well throughout the height of
the beam.
The Ioop outside that layer, the solve-beam loop, calculates the global strain state
needed to achieve the desired global load ratios (N, M, V).
The final outer loop, idle-dual, iterates on the shear strain profile until the
assumed profile matches the one calculated with the longitudinal stiffiiess method fiom
Chapter 6.
The Idle-Dual fbnction has the "idle" prefix as it is calleci during idle processing.
This allows the program to support rnultitasking on codperative multitasking operating
syaems. The first step in this function is to mate a new lord stage. With the initial
strains ffom above, one variable (generally curvature) is selected to be constant to
implement strain controlled behaviour. A cal1 is made to the solve-barn tiinction to
solve for the other 2 global strains that correspond to the correct moment: shear: axial
force ratios. The shear strain profile is then recalculated b d on the longitudinal
stiffiiess method. If the newly calculated profile is close to the assumed profile, the load
stage is completed. If the new profile is not close enough to the old one, then another
iteration is required. The longitudinal stifiess method calculates a shear stress profile,
which is converted to a strain profile using a secant stiffhess method for non-zero initial
stress levels and tangent method for zero initial shear stress levels.
The solve-beam fiinction solves for the global sectional strain state that causes the
desired load ratios. It first calls for the global tangent stifiess matrix as explained in
chapter 6, and then uses that to iteratively reduce the error in the sectional forces with a
tangent technique. The get-forces fiinction is the most important function called as it
updates the current sectional force state tiom a given sectional strain state.
The get-forces fnction will calculate the overall forces on a section based on the
given global strain state. For each node, a longitudinal strain is known as well as is the
shear strain. The transverse strain is calculated for each node to ensure that the node is in
equilibrium in the transverse direction. This is performed with the secant stiffiiess
method fiom Membrane-2000. The get-forces fnction first assumes that 8 layers
throughout the thickness are sufficient and these are calculated. The stresses are then
calculated at the quarter points of each layer and compared to the quadratic interpolation
from the existing nodes of the layer. If the quadratic interpolation matches the check
point, then the analysis will accept that layer. If the interpolation is poor, then the layer is
automatically divided in two.
These series of nested loops may sound intimidating, but in fact they mesh
together well and produce a fast solution for the problems.
7-4 Long Term Equations
Response-2000 includes a routine to allow automatic consideration of the effects
of shrinkage, creep and relaxation. These procedures are based on the ~ M I 3 ~ 0 - 9 4 ' ~
provisions and similar to the methods in Collins & ~itchell'. The reason to explicitly
include them is that the usually suggested method of deaiing with creep, increasing the
strain at which concrete cylinders mach peak stress, cari cause problems in an analysis for
shear. Using the method built into Response-2000 will avoid these problems.
A single load analysis is carried out by Respanse-2000, first ignoring al1 these
parameters, and at a load equal to the long-term moment value entered by the user.
m'hile shear is ignored to speed the calculation, axial bad is considered. This analysis
will result in a short-term longitudinal strain profile.
A new analysis is then calculated with the shrinkage applied, the prestressing
strands relaxed, and the concrete having a strain at peak stress modified by the creep
factor above. This will produce a second longitudinal strain profile wrresponding to the
long term behaviour of the beam at the sustained moment.
The difference between these two strain profiles is calculated and added to the
user defined shnnkage profile. This will then implicitly include creep and shrinkage in
the calculations. An analysis performed afier this will represent short-tem loading (i-e.
in a manner of hours to days) on a structure that has been loaded for a long term (Le.
many years)-
Note that the increase in concrete strengh that long term hydration will cause is
ignored as this is too dependent on individual mix properties.
The crack spacing is based on the CEB suggested crack spacing relationships:
Crack spacing = 2 c + 0.1 ddp (7-6)
Response-2000 and Shell-2000 calculate crack spacing over the depth of the
section, as it will change over the depth. For these calculations, the term c is taken as the
largest diagonal distance tiom the current depth to a reinforcing bar. The p term is taken
as the percentage of steel within a concrete area 7.5 4, above and below the bar. When
between different layers of reinforcement, the 0. lddp term is Iinearly interpolateci
between the calculated vafues at the bars. Ifa section is subjected to bending, the crack
spacing is not allowed to exceed the section depth.
For cases with no reinforcement, the crack spacing is selected as five times the
depth of the section.
7-6Hoop Reinforcement
Special treatment is given to transverse reinforcement that is in the form of hoops.
Figure 7-2 shows a column with hoops. It is assumed that there are no strains
perpendicular to the direction of shear ioading. This means that the part of the hoop that
is pointing that way will not experience any straining due to the shear. Confinement
would induce strain at this location, but this is not currently implemented in Response-
2000.
To account for this, the hoop strains c h are calculated fiom the transverse strains E(
based on a Mohr's circle:
= et sin2(a) (7-7)
Direction of assumed Shear Direction
zero $train
Shear Direction
Cross Section
-
Figure 7 2: Treatment of hoops in Rcspons+2000
Additionally, only the component of the force in the hoop in the transverse
direction rnay influence the shear response, that is, the hoop force must be multiplied by
the sine of angle alpha.
Response-2000 divides up the beam into 20 short segments. For each segment,
the axial load, moment, and shear force are determined fiom the applied loads. The
curvature and shear main associated with this load level is then interpolated fiom the
interaction diagram. This is integrated with the moment-area method to calculate the load
deflection relationship for the barn segment.
Shown in Fig. 7-3 is a Moment-Shear interaction diagram of a beam without
stirrups. The horizontal axis represents
moment and the vertical axis represents shear.
The outer line enclosing al1 the points
represents the failure envelope. Any force
combination that touches this line will result in
failure of the cross section. At the far right
hand side, the failure will be in positive
flemire, at the left it 4 1 1 be in negative
flexure, and the sloping top represents shear
-
Figure 7 3: Moment-sherir fai lures.
interaction diagram
Each of the squares within the
interaction diagram represents a solved combination of moment, shear and axial load.
This means that the cuwature, shear strain and longitudinal strain can be interpolated
throughout the interaction diagram using the values at the squares. The interpolation is
performed with finite element shape fiinctions.
If a line representing the shear and moment along the length of the beam is traced
on the interaction diagram, as shown, the curvature and shear strain can be calculated for
each point along the line and integrated together to predict the member load-deflection
curve.
A difficulty arises when using this method with shear, however. It is known that
due to diagonal cracks requiring horizontal projection, amongst other reasons, that it is
not appropriate to do an analysis with the fll shear and moment directly below the point
load or over the suppon. It is generally assumed, however, that it is appropriate ta do
analyses with fll shear a distance d or d, away fiom the point loads o r suppons. To
mode1 this the following mle was defined as shown in Figure 7-4 to define what parts of
the sectional loading is "active" for the deflection analysis.
It is proposed that within a distance of d fkom a point load and d fiom a support,
there are other support mechanisms that mean the entire load is not supported in a
sectional mode. These other modes inciude direct strut action fiom the load as well as
clamping fiom the load itself and fiom the support. Recall that Response-2000 assumes
that there is no overail stress in the transverse direction. Clearly that is not true directly
under the load nor directly over the support.
Based on this assumption of other mechanisms supporting the shear, the active
shear force diagram has been "clipped" over a distance d as shown in Fig 7-4.
I
A
8
C
Beam Loading ri 4 I r B LI t rI
Note that this is only safe if the "other mechanisms o f support" are indeed able to
resist the load. For simple beams with point loads or uniform loads, this is a good
assumption. It is consistent with the method used to calculate the experimental
verification for Response-2000 in Chapter 10, for example. In the event, however, the
load is supported in a peculiar way, Say with the beam hanging fiom the top flange rather
than sitting on a supports, this may not be a safe assumption.
Refemng to Fig. 7-4, Response-2000 does the anaiysis for half of the beam, fiom
A to B. For both the point load and the uniform loading, the familiar shear diagrams are
trimmed for a distance d fiom the ends.
In the event that it is desired not to trim the active shear diagrams, Response-2000
can be told t o do so. Selecting the left side support (location A) as a "hanging support"
results in the piece not being clipped fiom the diagram on the left. If the right side
support (location B) is changed to be a "hanging l o a d , the right side will not be clipped.
In the event that the regions where the shear is clipped interfere with each other,
that is, that the beam is less than 2 d long, Response-2000 assumes that the sectional
force demand is equal to the lower region that is still shaded. As this region does not
reach up to the top of the shear diagram, it is predicted that beams shorter than 2 d will
have increasing strengths. This is shown in an example in Chapter 10.
In the cornparison graphs, the thick solid line represents the MCFT,the thin solid
line represents the Rotating Angle-Softened Tmss modell', and the thin dashed line
represents the Fixed angle-Softened Tmss ~ o d e l l ' . If only one thin line is visible, then
the fixed and rotating angle models predict the same result. Al1 calculations were
performed with Membrane-2000.
For each graph, the horizontal axis is shear strain in parts per thousand, and the
vertical axis is shear stress in MPa.
8- 1- 1 Tests of Pang and H S U ~ O
These tests involved normal strength concrete subjected to pure shear. The A-
senes contained qua1 reinforcing ievels in X and Y directions and the B ~ n e contained
s
different levels in X and Y. Specimen A l is not included due to an edge failure.
O 5 10 15 20 O 2 4 6 8 10 12
Note the tendency for the very high strength concrete elements to be slightly over-
predicted by the MCFT. This is due to the compression soflening relationship and is the
source of the suggestion to use the ~orasz*'relationship for high strength concrete as
listed in Chapter 5.
9-1 General
As the work for this thesis was comrnencing, a number of shear tests on large
lightly reinforced concrete beams were being performed. They were suggesting that the
use of very high strength concrete 0 9 0 m a ) was not attaining the expected strength
increase over normal strength concrete, but instead, was providing a strength decrease "
40
. The crack faces were observed to be relatively smooth with cracks going straight
through aggregate rather than around the aggregate as with weaker concretes. This
apparent reduction in aggregate interlock called into question some of the assumptions
that had been made regularly in the analysis of high strength concrete. A small test
program using the shell eiement tester was developed for this thesis to answer some of
these questions. Another series of tests were also commenceci to attempt to find what
range of concrete strengths had this efTectm.
The Modified Compression Field Theory assumes that the majority of the t e n d e
straining in the concrete will effectively happen at a series of parallel cracks at an angle
theta. Compression is transferred parallel to this in the uncracked concrete between the
cracks. In elements reinforced with different arnounts of reinforcement in the X and Y
directions, the angle of cracks is found to rotate during the test as the weaker direction of
steel strains at a different rate than the strong direction. This means that the concrete
strut carrying the compression parallel to the current crack direction may in f a a have
cracks in it fiom previous loading. These earlier cracks would be at an angle to the
compression and would be subjected to shear stresses on the crack face that would be
resisted by aggregate interlock. Traditionally it was assumed that these earlier cracks
would be relatively small, and thus able to resist the applied compression. The question
was whether very high strength concrete specimens were able to achieve enough
resistance to cany sufficient compression. Earlier tests by hid de" on lightweight
concrete, which also tends to crack through the aggregate, found that the assumptions
about aggregate interlock were reasonable, but those tests had only about 1-2 MPa of
compression. Higher levels of compression would presumably be more critical and
potentially shed light on the unexpected beam test results.
To answer this question, two &Il-scale shell elements were designed and
constructed. Both were geometrically identical, but the fim used normal strength (34
MPa) concrete, and the second used very high strength (1 10 MPa) mncrete. Differences
between the behaviour of these two elements would provide information about the effcts
of using very high strength concrete.
The goal was to m t e the elements as sensitive as possible to the effcts that were
causing concem. As a result, the element thickness was large and wntaine
reinforcement with relatively poor crack control characteristics. Previous shell elements
had a cover of 10 mm, while these elements had 70 mm clear cover. The in-plane
spacing of the bars was kept at a reasonable spacing to ensure that the elements would act
in a uniform way across the test region. The reinforcement levels were seiected as
1.758% in the strong direction and 0.400% in the weak direction. This level of
reinforcement was selected to ensure that if the element achieved biaxial yield of both
directions of reinforcement, as expected, it would be subjected to signifiant
compression. Loading was applied in tension horizontally with the strong direction of
reinforcement angled at 32 degrees to this direction. This combination was selected to
achieve the brittle loading in the test region of biaxial tension and shear. The loading and
the reinforcement direction were predicted to induce large rotation of crack angles during
the tes&yet still maintain large principal compression. The normal strength element was
tested at an age of four months, but the high strength concrete specimen was tested at an
age of almost three years. This delayed testing had the desirable nsult of maximising
any shrinkage and dryng effects.
Overall, these two elements were unusual in that they were amongst the largest
elements tested in the shell element tester, with the highest strength, largest cover, largest
age, and were the first elements tested with reinforcement at this angle.
The details o f the specirnen geometry are shown in Table 9-1. Table 9-2 shows
the material properties. Figure 9-2 shows the stress-main characteristics of the
reinforcement. Figure 9-3 shows the reinforcing grid on the North side of the element,
looking north (Le. looking through the element). Table 9-3 defines the instrumentation
and loading of the element. Appendix B contains detailed Zurich surface strains and
photographs of the elements.
- 1176 -307
!- 3217 -1848 - 307 -1530- -1530 -
2
FRONT ELEVATION SlDE ELEVATION
-
Figure 9 1:The University of Toronto Shell Element Tester
Table 9-1 Specimen Geometry
Geometry : Square sheU element 1626x1626 mm.
Thickness: 398 mm
Clear concrete cover: 65 mm at connecter blocks
70-80 mm at centre of specimen (due to tying of bars)
X-Direction steel
No. SOM @ 86 mm both sides 1.758 %
Y-Directionsteel
No. 10M @ 126 mm both sides 0.400 %
2-Directio~steel
- +
4 6 mm bars for instrumentation -0.00 %
10M bars placed on outer side of specimen both sides for bad crack control
6 mm 4 bars hooked around 20M bars for good bond
-
Figure 9 2: Stress-strain properties of reinforcement
r t 1 BAR LIST
No. 20M
275 1 4
110 4
BAR LIST
No. 10M
I LENGm(mn) I NIMBER 1
Loading:
Specimen was connectai to loading yokes on al1 4 sides, though top and bottom yokes
not comected to jacks. 6 igid Links us& 3 in-plane (2 at botiom, 1 at East side), and 3 out of
plane (2 at top, 1 bottom centre). M n g in axial tension in horizontal direction.
Shear
F, = 0.716 Fh
Fy = 0.284 Fh
Axial
Fh v = 0.450 Fh
For this element, the failure was relatively ductile with ultimate failure being
controlled by rupture of a number of the 1OM bars and spalling of the cover. Table 9-4
describes individual test observations during the test, and Figure 9-5 shows the horizontal
deflection versus horizontal stress for the element. Table 9-5 summarises the
measurements made during the test.
-
Figure 9 4: Element HS1 mon aitcr cracking and just b e f o n failun
Table 9-4: Observations During Testing of HSl
Loading: Initiaily loaded up to 1 MPa tension, unloaded and then reloaded to "work outw
initial strains
L a d Stage 1 : 1 Crack on South side, 0.15 mm wide. No cracks on North side.
Loading: 2 audible swnds as specimen cracks over height, take a load stage
Load Stage 2: 3 cracks on N side, 3 cracks on S side.
Mark cracks in Black, miss one on Bottom South East corner.
Ming: Another audible crack, with large deflection. Take load sbge immediately
Load Stage 3 : Mark new cracks in black. Re-mark original cracks with red dotted line.
Many new cracks at a rotated angle. Large region uncracked on east side
Ming: Pause during tesbng to observe cracks, tben continue (LS 3B)
Laad Stage 4: Many new cracks, lower angle tban before, some ia order of 30-33 degrees fiom
vertical. Stop for the &y, take load down to 0.2 MPa
Mark new cracks in green.
-
Figure 9 5: Loading of HS1
0.00
o.m
4.01
4.m
4.m
4.01
4.m
4.m
oa?
4.01
4.01
0.00
0.m
0.01
0.m
0.01
0.a
o.m
O.m
0.03
0.05
0.-
0.07
0.-
0.-
0.11
0.1 1
0.12
0.13
0.14
0.15
0.16
0.10
0.0
0.1)
02s
ors
011
o s
O24
0.a
O31
O20
030
0.18
0.1)
O20
Ozl
oz2
02
023
O24
024
024
025
011
al,
0 3
0.40
0.0
0.61
0.64
0.87
0.71
0.75
o.n
0.0
0.82
0.0
0.78
1.m
0.77
0.n
0.7s
9-4 General Observations: HS2
Element HS2 acted similarly to HS 1. Figure 9-6 shows photos of the element
soon &er cracking and near failure. Due to the higher strength concrete, the stress at
first cracking was higher. The initial cracks in HS 1 were 0.15 mm in width. For HS2, the
initial cracks were 0.35 mm wide. The extent of cracking at failure was similar. Crack
slips were measured for this element and found to be about the same as the width of the
cracks. That is, a 7 mm crack was found to have slipped about 7 mm in the direction of
the crack. Table 9-6 lists observations made during the testing, and Table 9-7
summarises the measurements made during the test.
Rupture of many
bars 1
O 5 1O 15 20 25 30 35 40 45
Horizontal Stmin (nnJm)
-
Figure 9 6: Loading of HS2
-
Figure 9 7: Elcmcnt ES2 soon rfter cracking and just before failure
Table 9-7 Element Summary: HS2
105 0%
127 1.02
151 138
1% 2t3
a9 29
232 267
250 r#r
a3 3.01
269 zm
Xl 240
312 2Q
325 2-44
Mo 2.25
U6 1.49
351 la
se 0.U
4w Off
417 030
42s O17
434 0.18
46 1 4.0s
470 4.17
476 422
482 424
522 42s
566 4s
560 426
605 425
633 435
661 434
8.37 423
sa1 4-15
919 4.0s
SU 0.01
95 1 0.01
972 0.07
94( 0.11
1023 0.1s
llnl 0.10
1086 020
1100 0.19
f 165 0.02
1179 0.10
1192 0.14
1214 0.12
1234 4.01
in1 5.12
ln9 0.01
1759 0.03
lm 0.a
1781 0.00
176 0.06
1787 0.03
1828 4.10
lm2 5.10
:FM5 0.42
1912 417
19% 4.26
1947 4.13
l m 4.61
1977
1sw 432
l
W 4.44
1%7 4.67
1
w 4.
1991 4.63
1 993 -0.46
4.14
osi 2.m O 40
9-5 Cornparison of behaviour and predictions: AS1
The following plots show the behaviour of the element. Figure 9-8 shows the
shear-shear strain plot for the element. Also plotted is the prediction fkom the MCFT and
the Rotating Angle Softened Tmss model. It can be seen in this figure and the three that
follow that the MCFT prediction for this normal strength concrete is good, though it
overestimates the stress levels over a large part of the test. The failure load was well
predicted. The RA-STM prediction is relatively poor in cornparison for both strength and
ductility
25 1 MCFT-1987 Prediction
10 20 30
Shear Strain ( M m )
- -
Figure 9 8: HSI Shear stress shear strain
Figure 9-9 shows the principal stress and strain angle along with the predicted
angle theta fiom the MCFT. Note that the angle of stress and strain are not equal, though
assumed so by the MCFT. The calculated angle is a reasonable approximation of both,
however.
Predicted Angle (MCFT)
Anale of
N Principal Strain
r y i aor
Principal Stress
-
Figure 9 9: HS1 Angles
Figure 9-10 shows the principal tensile stress-strain relationship obseved with the
predicted retationship. The fit is quite good considering that tension measurements are
-
Figure 9 10: HS1 Tension stiffening
small differences between large numbers and hence tend to have high scatter. The
measured values d o not extend d o m to zero with higher arain suggesting that some
other mechanism was arengthening the panel, such as kinking of the reinforcement at a
crack. The loop where the curve goes negative was during the overnight unloading of the
panel.
Figure 9-1 1 shows the measured compressive stress-main relationship and the
predicted values. The results, again, are good. Note that the observed stresses and strains
were drarnatically lower than the shown cylinder curve. The concrete failed at a stress of
only 20 % o f the cylinder strength. The maximum compressive strain observed was
about twice the strain at peak stress for the cylinder.
-
Figure 9 11: HS1 Compressive response
9-6 Cornparison of behaviour and Predictions: HS2
Figure 9-12 shows the shear stress-shear strain plot for the element. Also plotted
is the prediction fiom the MCFT and the Rotating Angle Softened Tmss model. Recall
that this element was made of very high strength concrete and was tested partly to
determine if the MCFT is unconservative for such elements. It is seen that the basic
predictions are indeed unconservative for this element. The strength is substantially
overpredicted, as is the ductility. The prediction here was made with the Porasz/Collins
concrete model, as recommended for very high strength concrete in chapter 5, yet was
still unconservative. Note that despite being calibrateci partly on large high strength
elements, the RA-STM also does a poor job and is almost equally unconservative as the
MCFT for this element.
MCFT Predicb'on
1 t
O 10 20 30 40 50 60 70 80 90
Shear Strain (mmlm)
-
Figure 9 12: HS2 Shear stress-shear strain
Figure 9-13 shows the principal angles. The MCFT prediction of the angle is not
as good as it was for HS 1. The angles of principal stress and strain are similar in the
tests, as before, but the predicted angle is 10-20 degrees away fiom that.
MCFT Predicb'on
Angle of
Principal Stress
15 25 35 45 55 65
Angle (deg)
-
Figure 9 13: HS2 Angles
Figure 9-14 shows the observed tensile stress-strain response fiom the test. As
with HS 1, the prediction is quite good. Note that the MCFT assumption of zero tension
stiffening aAer yield of the steel is well supponed by the data
Ercperimental C m
70
-
Figure 9 14: HS2 Tension stiffening
Figure 9-15 shows the observed compressive stress-strain curve for test HS2. The
data shows virtually no strain development on loading as a number of major cracks
intersected only some of the LVDTs used to calculate the strains. The maximum stress
observed is basically the same as for the rnuch weaker wncrete of HS 1. Note that more
stress was observed in the concrete than predicted, but the result was still unwnservative
as the angles were poorly modelled.
tu
b
3
cn
g
3i
O
>
cn
e
-lu
O 10 20 30 40 50 60
Shear Strain (mmlm)
-
Figure 9 16: HSl and HS2 shear-shear strain
hardening
without stmin
hardering
O 20 40 60 80
Shear Strain (mmlm)
-
Figure 9 17: HS2 Predictions without strain hardening
Chapter 10: Experimental Verification of Response-2000
As a beam~columnsectional analysis program that includes shear, Response-2000
will be illustrateci using selected experimental results for beams and columns tested in
shear over the past 40 years. Chapter 12 provides additional information on how to
perform such shear analyses with Response-2000.
The first part of this chapter demonstrates the value of the proposed changes in
tension stiffening relationships as explained in Chapter 5.
The third part of this chapter shows Response-2000 compared to a large database
of 534 beams. There have been over 10,000 shear tests reported in the literature over the
past 100 years. When databases that aggregate al1 this data, as are currently being
prepared, are available, the program can be compared to more tests. Tests were
originally selected for this data set simply in terms of what was available at hand. None
of these tests were discarded if the prograin predicted them poorly . Later, a more
systernatic procedure was chosen to select additional elements by examining what
regions of input parameters were missing fiom the existing database. There is a strong
bias in the selected data towards larger members. Before about 1965, beams tested in
shear were generally about 12 inches deep and 10 feet long. This was based on storage
restrictions, loading restrictions and tradition. Since that time, it has been found that
larger beams tend to fail at a lower shear stress than geometrically similar smaller beams.
As these larger beams tend to be less conservatively predicted, they were preferentially
selected. The average beam depth in the database is only 500 mm, however, as there are
a number of important tests that were included fkom the older database.
Figure 10-1 shows the experimental over predicted s h w strength for these beams
if the traditional tension stiffening relationship of the MCFT, as shown in the figure, is
used. The horizontal axis has been selected as overall beam depth because this was found
-
Figure 10 1: ExperimtntaVPrtdictcd, Base MCF
to have the strongest correlation with the data. It cm be seen that there is a strong
correlation with larger members being predicted poorly.
Figure 10-2 shows the same set o f data as predicted by Response-2000 using the
tension stiffening relationships fiom Chapter 5 . It can be seen that the prediction is much
better.
d
?
O
t
cn
f
Ca
1-2
h&-l 1,.
+
3---- +
T V - - - - -
e
0
8
0.8 - a
9
lf 0.4
-
- Tension stiffening
-E 0.2 - described in Equation 5 9 , Chapter 5
x
W
O , I 1
-
Figure 10 2: Experimentanredicted, Response-2000
The remainder of this chapter shows the predictions of Response-2000 with the
newer tension stiffening relations as proposed in Chapter 5 .
10-2 Arbesrnan and Conti: Prediction of sectional response
Perhaps the most important beam test in the development of the Compression
Field Theory was Bearn CF1 (Compression Field 1 ) tested by Arbesrnan and in
hollow box girder beam, heavily insuumented. It was loaded in such a way that the test
region was subjected to negative moment near the person in the photograph and positive
moment just to the right by the dia1 gauge. There was a point of zero moment in the
rniddle of the test region. Failure of the beam was by flexural yielding at the positive
moment location and crushing of the web of the box girder (shear failure) near the
location of zero moment.
The bearn was entered into Response-2000 as an 1-beam section. Figure 10-3
shows the predicted and experimental moment-curvature results at a location below the
point load. The experimental curvature was detennined from measured longitudinal
strains of the top and bottom steel. The analysis was performed without shear (Le. a
flexural analysis.) The results obtained are remarkably go&. Unfominately, this
extremely high level of precision is only ofcasionally attained with Response-2000, or
with any other analysis procedure for reinforced concrete.
-
Figure 10 3: Moment-curvature of CF1
Figure 10-4 shows the shear force versus shear strain for CF1 in the middle of the
test region where the moment was equal to zero. The shear strain was measured by
taking the difference in strains of two diagonals at 45O to the horizontal. The very good
-
Figure 10 4: Shur-shur strain CF1
agreement between theory and experiment shown in the figure is more representative of
what can be expected for Response-2000.
-
Figure 10 5: Effect o f aggregrte sizt
10-4 Shioya et al: Size effect in shear
In what is surely the most impressive series of shear tests yet performed. a large-
scde test program in ~ a ~ a n ' 'tested
. ~ ~ beams thnt varied in depth from 1 0 0 mm (4 inches)
up co 3000 mm ( 10 feet). The largest of these bearns was 36 m ( 120 feet) long. weighed
almost 300 tonnes and contained 1O times the volume of concrete of a standard bearn
shear test. The testing of the largest of these beams is shown below.
These tests mode1 a foundation style structure. Such structures generally have light
amounts of longitudinal reinforcement and no transverse reinforcement. It might be
felt that a structure 3 metres deep is unrepresentative of real construction practice. This is
not the case however. Many multi-storey buildings contain bearns that are more than 1
metre deep, ofien transfer girders above the first floor of a building. A typical industnal
building near Toronto contains 2 metre deep girden supporting process equipment
weighing thousands of kilonewtons. In portions of the Toronto subway system tunnels,
one-way slabs are 3 metres thick without transverse reinforcement. Similar 3 metre deep
tunnel-roof slabs exist in Japan over sub-surfacehighways, where the prevalence of
earthquakes adds a level of concem. The Petronas Towers in Kuala Lumpur Malaysia,
currently the tallest buildings in the world, are built on 2-way slabs that are 4.5 metres
deep of solid concrete. Finally, in Japan there are 100 m diameter underground liquid
natural gas storage facilities that contain slabs at their base designed against groundwater
pressure. These base slabs are up to 8 metres (25 feet) thick of solid concrete without
transverse reidorcement. Each of these examples is already built in the field, sometimes
using design codes that do not deal with the size effect in shear.
Figure 10-6 shows the predictions of Response-2000 to the test series that
-
Figure 10 6: Shioya et al size effect
included the 3 metre deep beam above. The trend is remarkably well predicted by the
program.
The ACI code assumes that members without stirrups can resist a stress of
0.167,c (h.IPa units, 2& psi units) prior to failure. This was based upon laboratory
testing which primarily used beams about 300 mrn deep. Note that the value of 0.167 is
reasonable value for elements of that depth based on the graph above. On the other hand,
the largest of the beams failed at about 45 % of this value. This is an unsafe situation.
Response-2000 is also able to predict the total load deflection response for
prismatic beams. The calculated response for the largest of the Japanese beams is shown
compared to the measured response in Fig 10-7. It can be seen that excellent agnement
was obtained.
0 1 I 1 T I
0 20 40 60 80 100 t 20
Midspan Dtfbctiorr (mm)
-
Figure 10 7: Load deflection, 3000 mm deep beam
The following figure shows t h e observed crack pattern observed by the Japanese
researchers as well as the deflection prediction page from Response-2000. The predicted
crack pattern is shown for half the beam and is reasonably accurate. The distribution of
cuwature and shear strains shows the strong interaction which occurs in this beam
between shear and moment. For exarnple, it might be expected that the highest shear
strain would occur at the end of the beam where the shear is highest but this is not the
case. Rather, it occurs in a region fiom about 1/6* of the span to 113" of the span where
both the shear and moment have high values.
Out of this extensive selection of data, a single series is shown here. These tests
show the effect of shear span to depth ratio (dd ratio) on the shear strength of 6 10 mm
deep beams with 2.8 % longitudinal reinforcement and normal strength concrete.
Response-2000 predicts the data very well for beams with an d d ratio greater than 2.5.
Beams shorter than that are aff'ed by direct stnit action supporting the load. The
dashed Iine shows the results of calculations done with a simple moment to shear ratio
style calculation as for the tests with an a/d greater than 2.5. This is seen to be very
conservative for these short beams. The method proposed in Chapter 7, shown by the
thicker line, does a much better job at predicting the shear capacity. This technique
calculates the strength, and deflection, from the moment-shear interaction diagram.
-
Figure 10 8: Effect of a/d ratio
10-6 Moody, Viest, Elstner, Hopestad: Concrete strength: small beams
In an important senes of tests used to calibrate the benchmark 1963 AC1 code,
research4 was carried out on bearns without transverse reinforcement but with a large
amount of flexural reinforcement against different concrete strengths. Data like this was
the basis of the code expression of 2Jf;E (psi units) for the V, t e m in the wde as
mentioned above. The graph shows that Response-2000 is able to prdict this type of
behaviour well.
For simplicity, the authors of the AC1 code chose to select a lower Iirnit on the
then existing test data and use this to specify a safe level of loading for a structure. In the
figure, the AC1 code is seen to be conservative for these tests.
-
Figure 10 9: Effcct of concrete strcngth on small heavily reinforccd beams
10-7 Angelakos: Concrete strength: Iarge lightly reinforced beams
o s ~ completed an experimental project at the University of
~ n ~ e l a k rec~ntly
Toronto that studied the variation in shear strengths of large lightly reinforced concrete
beams made with different strengths ofconcrete. Like the srna14 huvily reinforced
beams in Fig. 10-9, these bearns had no transverse reinforcement. The small beam tests
would suggest that the AC1 code shear strength equation should be conservative for the
beams up to at least 50 MPa.
The test results for the large lightly reinforced beams are shown in Fig. 10-10. Tt
was found that the AC1 code for these members is unconservative over the entire range of
concrete strengths. The beam with the highest concrete strength failed at a shear less than
half the predicted capacity of the AC1 code. Response-2000 can be seen to do a better
job of predicting these strengths, though it too becomes rather unconservative for the
higher concrete strengths. Further research is required to explain this discrepancy, with
work of ~ u ~ t a looking
*' promising.
O ! I I I 13001
T I T , I r 1 1
20 30 40 50 60 70 80
Concrete Striength (MPa)
-
Figure 10 10: Effect OC concretc strength on large barns without stimps
10-8 Adebar & Collins: Effect of axial tension
For members without stimps, the AC1 code suggests that the shear strength is
very sensitive to axial tension, reducing to zero for an axial t e n d e stress of 500 psi.
Figure 9- 10 compares the AC1 and Response-2000 predictions with the expenmental
results reported by Adebar and ~ o l l i n s ' ~It. can be seen that the AC1 predictions are very
conservative for members with high tensions.
-
Figure 10 11: Effect OC axial tension
10-9 Khalifa: Transverse Reinforcement on round columns - - --
The examples presented so far were 6om rectangular beams or box bearns.
Khalifa4' tested round colurnns with a small axial compression in a specially designed
testing fiame. The primary variable of the study was the level of transverse
reinforcement. It cari be seen that the increase in shear strength from increasing
transverse reinforcement is predicted accurately.
1.4 0
O
1.2 OwO O 00
O O
Bs O
-----
1
77------ 3-
gg
-g 'Po
&r 0.8
0.6
O
O O
-E 0.4
a
W 0.2
C
O I L i
O 2 4 6 8 10
Drape Angle (degrees)
- -
Figure 10 13: MacGregor Drape of Reinforcernent
Figure 10-14 shows the quality ofthe predictions of Response-2000 versus the
percentage of transverse reinforcement for the tests reported by MacGregor. It can be
seen that the predictions are go04 but for low levels of transverse steel, Response-2000
becomes more conservative. Recall that in Chapter 1 it was shown that prestressed
concrete beams may be more conservative than reinforced wncrete beams due to the
presence of transverse compression in the web of the beams. This is a possible
explanation for the somewhat wnservative nature of the predictions made for these tests.
-
Figure 10 14: Macgrcgor, emect of transverse reinforcement
10-11 Benzoni, Priestley and Seible: Interloeking spiral column
A reIatively new type of bridge pier used in California involves the use of two
The cycIic loading for this column was as shown below left. Note that the axial
load in the column was different for the push and
Loading Configuration
-
H -
I i
/777/////.////////777777777/7/////rn
Vertical Load P:
P= 994 + 2.45 H (kN) when H pushes
-
P= 994 4.33 H (kN)when H pulls
pull cycle in the loading. A photo of the beam at failure is d s o shown above. It can be
seen that the failure involved longitudinal cnishing bands with no evidence of flexural
failure.
An analysis was carried out a distance d, 60m the bottom of the column with
Response-2000 including the effea of shear. The results of this are shown here:
Obsewe the predicted cause of failure fiom the Response-2000 analysis: local
cnishing of the web and high shear strains and high transverse strains just above the
location where the lower hoop ends. That is, about 75 mm up fiom the mid-height of the
section. In the bottom left of the 9 plots, the maximum allowable compression is
approaching the applied stress. At higher deformations, the concrete crushes and the load
reduces. While there is longitudinal yieIding, there is still plenty of flexural capacity in
the non-yielding bars. This local cnishing of the web is symptomatic of a longitudinat
shear failure. It can be seen in the photo of the failed specimen that the concrete failed by
crushing at about the same place as predicted by the analysis, over a substantial height of
the column.
4501
Dbpimceiimt (mm)
-
Figure 10 15: Inter4 predicted load delection
151
10-12 Cornpanson to 534 Beams
The previous pages demonstrated the quality of predictions of Response-2000
against individual test series. The following pages demonstrate the quality of the
predictions against a database of 534 beams. This database includes rectangular beams
and columns, round colurnns, prestressed sections and 1bearns. Table 10-1 summarises
the tests and Appendix C lists al1 the data. The performance of Response-2000 is shown
below compared to different variables fiom this database.
-
Figure 10 16: ri/d ratio
10- 12-2 Beam depth
Figure 10-17 shows the results compared to the depth of the beam. It can b seen
that the size effect in shear is successf1ly accounted for.
-
Figure 10 17: Beam depth
Table 10-1 Summary of Response-2000 Experimental Verification
O b 1.3 MP.
Ot012MPi
~halifa~' O to 2.7 M P i
Aregarni'
Elnnaty, N i l m O b 4.8 MP.
and lat te^
Podgomiak- O to 0.35 MP.
~taniP
Shioya et aez* O
Pasley, Gogoi,
Darwin
McCabem
Plaskas O to 0.77 MP.
Darwin"'
Ozcebe, E ~ a y O to 0.71 M P i
and Tankutw
Roller, usae el^ O to 8.1 M P I
O to 2.0 M P i
1O-12-3 Concrete strength
Conceni has been expressed about the accuracy of code equations for the shear
strength of members made fiom very high strength concrete. Figure 10-18 shows that
Response-2000 is able to account for concrete strengths well.
-
Figure 10 18: Concrete strength
10- 12-4 Longitudinal percentage of reuiforcement
Beams with a lower percentage of longitudinal reinforcement have been observeci
to fail at lower shear strengths. Figure 10-19 shows that Response-2000 is able to predict
such behaviour well.
O 1 2 3 4 5 6 7
Longitudinal Percentage of Reinforcement (*A)
-
Figure 10 19: Longitudinal Percentage o f Rcinforccment
10-12-5 Transverse percentage of reinforcement
Figure 10-20 shows the ability of Response-2000 to predict shear failures as a
hnction of how much transverse steel is provided.
-
Figure 10 20: Transverse percentage of reinforcement
10-12-6 Shear strength
Figure 10-21 shows that Response-2000 is not biased towards stronger or weaker
specimens. Note that the horizontal axis in this case is logarithmic.
-
Figure 10 21: Shear strength
10- 12-7 Overall Predictions
Figure 10-22 shows the entire dataset compared as experimental shear vs.
predicted shear capacities. The average experirnental over predicted shear strength for
the data set is 1.05 with a coefficient of variation of 12.00/0. Figure 10-23 shows the
predictions of the AC1 code in a similar format. The AC1 code equations yield an
average of 1.20 with a coemcient of variation of 32.1%. Note that while there is a
signifiant scatter to the AC1 predictions, the predictions are al1 conservative for failure
shears less than 70 kN (15 kips). This is the general range of failure shear strengths for
small tests such as those shown in Fig. 10-9. It would appear that the extrapolation of
these results to larger, and hence stronger beams, is not well modelled by the code.
While Response-2000 compares very well to the code provisions for predicting
strength, it is important to redise that the results fiom Reponse-2000 have an important
difference over most code methods. The AC1 provisions, for example, were based on a
curve fit to a large data set. As such, it should be expected that such a method would do
well at predicting shear strengths. For Response-2000, on the other hand, the majority of
the constitutive methods are based on a totally different kind of experiment, the shear
panel experiment. While it may be assumed that these panel tests would be directly
applicable t o beams, indeed that is why the shear tests were done in the first place, it is
satisGing to see that they are.
-
Figure 10 22: Responsc-2000 experimental and predicted shear strength
n = 442
Average = 1.20
C.O.V. = 32.2%
-
Figure 10 23: AC1 experimental and predicted shear strength
0.45 0.6 0.75 0.9 1.05 1.2 1 1.5 1 . 1.8 1-95
Btperimenta 1 Predldrd Shear Strength
-
Figure IO 24: Response-2000 expenmtntriypredicted ratio
0.45 0.6 0.75 0.9 1.05 1.2 1.35 1.5 1.65 1.8 1-95
Rperlmentd 1 Predlded Sherr Strength
-
Figure 10 25: AC1 experimentaVprtdicted shcar strength
Chapter 11:Experimental Verification of Shell-2000
SheI1-2000 will calculate load-defornation relationships and strengths for plates
and shells. In a sense it is a superset of Response-2000 for beams and Membrane-2000
for membranes. The program will be compared to these other p m m m s as well as
experimental evidence.
Experiment
O 5 10 15 20
Shear Strain ( MN
-
Figure 11 1: SE6 In plane shur strcngth
Kirschner also tesied a series of similar panels to produce an interaction diagram
ofin-plane shear stress and flexure. Each element had 2.92 % total reinforcement in the
X direction and 0.98 % in the Y direction, 40 MPa concrete and was 285 mm thick.
Figure 11-2 shows the predictions of Shell-2000 compared to these tests. Note the
inclusion of the pure shear strength predicted by Membrane-2000 as well as the
prediction fiom Response-2000 prediction of the pure flexural strength. It can be seen
that the predictions including strain hardening are good, but the predictions ignoring
strain hardening are rather conservative for the cases with higher moment. A Response-
2000 analysis with no strain hardening shows the sarne patterns as Shell-2000 does.
Membrane-2000
Pure Shear
Shdl-2000 *th
strain hardening
No strain
Hardenina
-
Figure 11 2: M-V, interaction diagram
For a more detailed look at one of these elements, the load deformation of
element SE4 is compared to the predictions of Shell-2000 in figures 11-4 t o 11-9. This
element was loaded with 500 k N d m of moment for every 1000 kN/m of in-plane shear.
The pre-yield predictions are good, but Shell-2000 is underestimating the aiffness afier
yield. Note in Figure 11-8 that Shell-2000 is predicting significant y direction curvature,
but the tests measured virtually none. This and the over-prediction of twisting may be
related.
= 04- - - - 7 - - - - 7 - - - + - - 1
O 2 4 6 8 10 O 1 2 3 4 5
crs t n n (mmlm) c, sbrln (mmlm)
Shell-2000
Ultirnate strength
-20 -1 5 -10 -5 O 5 10 15 20
In Plane Sheor (MPa)
predicted and measured effect of in-plane shear on the out of plane shear strength. Note
that on the nght side of the figure, the in-plane shear adds compression in the direction of
the out-of-plane shear, increasing the strength beyond that of the zero in-plane shear case.
The reverse is true for the lefi side of the diagram. The solid squares in the figure are the
maximum loads resisted, but some of the elements were noted to have been restrained by
the testing apparatus, partly explaining the degree of conservatism on the lefi side. For
cornparison, the out of plane shear stress at first yield of the stirrups is also shown.
As a final cornparison of the out of plane shear abilities of Shell-2000, the
predictions of Shell-2000 will be compared to Response-2000. A shell of reinforced
concrete identical in the X and Y directions, sirnilar to that tested above by Adebar, was
entered into Shell-2000. A Response-2000 input file was made of the X direction of the
shell element as well. Moment-shear interaction diagrams were then made with Shell-
2000 of the shell elcrnent tested with moment and shear on the X face. Response-2000
was also used to calculate uich an interaction diagram. Finally, an analysis was made of
the shell element tested with the moment and shear applied at 45" to the reinforcement.
As this element is isotropically reinforced, it may be expected that there would be no
di fference between these t hree results.
Figure 11-1 1 shows the result of these calculations. As the Response-2000 and X
direction Shell-2000 analysis were modelling the same thing, it is satisfying that the same
curves are produced fiom the two prograrns. Of more interest is that the analysis done on
the element at 45" did not produce the same curve. The flexural strengths matched well,
but the predicted shear strength is about 85 percent of the value predicted for the direct
loading. This appears to be caused by the 45" case having to carry the shear over wider
cracks. The failure of this element was controlled by shear on the crack.
O 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 7 w
Moment (kflm)
Figure 11-11 M-V interaction diagram
Chapter 12: Analysis guidelines and examples
This chapter provides guidelines on how to use the program Responx-2000 to
predict the behaviour of beams and columns as well as some examples.
For the given example, then, the first analysis would be performed with an
incremental M=l .O and V= in the Response-2000 loads dialog box. That moment
capacity would then generally be mnverted t o an equivalent shear necessary t o cause the
moment. Then an analysis with shear would be performed with M=1.8 H m , V=1.0 kN
if the value of d for this beam was 900 mm. Note that these ratios are unit dependent and
would be M=5.91 ft-kips, V= 1.0 kip in US units.
Note that while the shear failure is calculated to happen at location S, it will in
fact happen with the opening of diagonal cracks as s h o w on the right of Fig. 12-1. A
consequence of this is that if there are stimps that Vary in spacing along the length of the
beam, it is appropriate to calculate the average amount of stirrups within a band of length
d in the beam centred on location S. For the beam above, the appropriate amount of
stirrups to add to the Response-2000 file would be the average level of stirmps within a
band stretching fiom 1-35 metres to 2.25 metres fiom the left support. Similarly, if the
stirrup spacing is very high, say p a t e r than 0.75 d, it may be necessary to assume a
reduced effectiveness for the stimps.
Consider also what would happen if the distance fiom the load to the support was
less than two times the effective depth of the member. In that case, the line AF would not
fit into the shear span, suggesting that a normal shear crack would also not fit. In cases
like this, Response-2000 will be very consewative, as the behaviour will be that of a deep
beam rather than that of a long beam. Chapter 7 explains how the use of the load-
defiection option gives a simple way to account k r this in Response-2000.
With pretensioned beams, one also needs to be mncerned about the bond of the
strands at the end of the beam. This is an area cunently k i n g researched, but a tentative
proposai has been developed for Response-2000. An analysis shouid be performed at a
location dl2 fiom the end of the beam (which assumes slip of strands and steep cracks).
The moment and shear at that location should come fiom the statics of the bearn loading
and be increased proportionally. The maximum stress in the strand can be estimateci as
the stress resulting tiom 750 psi (5 MPa) of bond stress on the strand circumference
aiong the strand fiom the end of the beam up to the inside edge of the bearing plate. This
stress should be induced in the strands with an appropriate prestrain, and the stress-strain
curve of the strands should be modified to provide a maximum at this level of stress.
While this suggested procedure needs further study, it does give results in good
agreement for the tests as shown in Chapter 10.
Using a simple dynamic analysis with a few assumptions, it was estimated that the
12 metre column would have an inflection point about 8 metres above fkom the ground.
With this assumption and an assumed axial compression of 14,000 W, Figure 12-3 was
generated to represent the load-deflection pushover results for this colurnn. The analysis
took about one minute. Once calculated, effects o f other assumptions of the location of
Load-Max Deflection
Predcted Failue
Location
As noted in Chapter 10, structures have already been built that are up to 8 mettes
in depth, generally footings. While it would be very difficult to perforrn an experiment of
that scale, Response-2000 has no dificulty predicting the behaviour of such rnembers.
The figure below shows the predicted shear strengths of large slab like structures
subjected to uniformly distributed load. The vertical axis shows the predicted shear
strength divided by the AC1 code predicted shear strength as a percentage. It is predicted
that siabs about 300 mm deep, such as the beams in Figure 10-9 of the 1st chapter, will
be conservatively predicted, that is a percentage greater than 100. On the other hand,
slabs 4 metres thick with, say, 0.5% of longitudinal reinforcement, are predicted to fail in
shear at a shear stress less than 50% of the AC1 code strength.
I I
.r
a
4,
1-4
1.2 +
T
Response-2000 Mnimurn Stimps
4
-
ACi stimps
htm
http://~~~.ecfutoronto.~a/-bentz/m2k. Membrane-2000
htm
h t t ://www.ecfutoronto.ca/-bentzIr2k.
~
htm
http://~~~.ecfutoronto.cal-bentzlt2k.
With cornputers being as fast as they are today, a typical Response-2000 analysis
taking less than 10 seconds on an inexpensive 1999-vintage computer for example, it
seerns reasonable to suggest that the time may have corne to allow engineers to use
experimentally verified tools more integrally in the design and analysis of structures. It is
suggested that the code could be changed to directly allow the use of numerical tools that
meet a number of requirements. These requirements would be that the tools:
Such tools would then be able to be used by engineers on any analysis problem, including
those that are poorly predicted today. It is suggested that the prograrns in this thesis
represent a good start towards meeting these goals.
Chapter 14: Areas of Future Work
While the four programs presented in this thesis are seen as being a good start on
the provision of a widely available set of non-linear sectional analysis tools, there are a
number of things that can be done to improve them. Some of these are summarised here:
Introduction
This manual covers the details of operation of the following programs:
Each of these programs is a non-linear sectional analysis program for the analysis
of reinforced concrete elements subjected to shear based on the Modified Compression
Field ~ h e o r ~ These
'. programs were written over the years 1996-1999 by Evan B e n e
P h .candidate at the University of Toronto under the supervision of Professor M. P.
Collins. Together they represent over 150,000 lines of C++.
The following guiding principles were used in designing these applications. They
were to allow fast checking for errors in input and fast interpretation of results with
ample graphies. They were to provide stable, state-of-the-art analysis techniques and,
finally, they were designed to leave the user knowuig more about the real behaviour of
concrete rather than less, as some computer programs seem t o do.
Each of the programs has a similar "look and feel" and has been designed to be as
intuitive as possible. This manual acts as an explicit explanation of what the programs
can do and how to make them do it. This manual does not attempt to provide any o f the
background into the analysis techniques used, as this is the main body of this thesis2.
Section 1of the manual provides a "quick start" type of description of how to
make simple input files for each of the programs as well as how to interpret the results.
Section II follows with a more detailed description of creating input geometry for
each program.
Section I defines the ioading options.
Section IV explains the types of analysis possible and explains the output fkom
each main screen of the program.
kU2VUAL Page A-3
Membrane-2000 Response-2000 Triax-2000 Shell-2000
Section V provides a description of some of the more advanced options that allow
customisation of the program.
These programs are available for no charge fiom the World Wide Web at the
following addresses:
htm
htt~://ww~.edlutoronto.ca/~entz/Rk. Response-2000
ecf utoronto.cd-benWm2 k.htm
h tt~://www- Membrane-2000
htm
http://www.edlutoronto.ca/-benWt2k. Triax-2000
htm
htt~://www.ecfutoronto.ca/-benWs2k. She11-2000
For trther details or for help in using the programs, contact the author at:
Evan Bentz
Department of Civil Engineering
University of Toronto
Toronto, Ontario Canada
M5S 1A4
benteecf utorontoxa
h44hWAt Page A-5
Membrane-2000 Response-2000 Triax-2000 Shell-2000
SECTION N: Analysis and Interpretation ................................................................ A-38
4- 1 General Information......................................................................................... A-38
4-2 Types of Analyses............................................................................................ A 4 0
4-3 Membrane-2000............................................................................................... A 4 1
Membrane-2000: 9 Plots General ........................................................................ A 4 1
Membrane-2000: 9 Plots Mohr's Circle .............................................................. A 4 4
4-4 Response-2000 ................................................................................................ A 4 8
Response-2000 9 Plots General........................................................................... A 4 8
Response-2000 9 Plots Cracking......................................................................... A-51
Response-2000 9 Plots Reinforcement ................................................................ A-51
Response-2000 9 Plots No Shear ........................................................................ A-53
Response-2000 Load Deformation Plots ............................................................. A-54
Other Load-Deformation Plots ............................................................................ A-56
Response2000 Full Member Plots...................................................................... A-57
4-5 Triax-2000 ....................................................................................................... A-59
Triax-2000 9 Plots General ................................................................................. A-59
Triax-2000 Other 9-plot Views ......................................................... .................. A-61
4-6 Shell-2000 .......................................................................................................A-63
Shell-2000 9 plots General .................................................................................. A-63
Shell-2000 9-Plot Views ........... ...................... ................................................... A-65
Shell-2000 Interpreting Crack Diagrams ............................................................. A-67
Shell-2000 Load-Deformation Plots.................................................................... A-68
Response-2000 allows only one cross section to be input at the same t h e . The
other programs al1 allow more than one with a catalog of elements available to select
between them. See section 2-7 for a description of the catalog.
Al1 programs allow the units to be changed at any time during the m i n g of the
prograrn using the "Options 1 Preferences" menu.
Re~ponse-2000Wizmd
The wizard for Response-2000 is more complex and is explained here in some
detail. There are four pages to this wizard with each shown and explained below.
Further down in the list are standard sections including CPCI-I beams, CPCI-box
beams, PCI Double-T's, PCI Single-T's AASHTO highway girders, and Washington
DOT sections. For these standard sections, the right entry fields are used for the
following four purposes. Fust to select a type fiom the selected category (pressing any
key will bnng up a list) and second to define a haunch (distance fiom the top of the
precast beam to the bottom of the slab). The third box defines the slab depth and the
fourth defines the effective slab width. Note that the slab width should be the effective
width for the purposes of analysis, rather than the simple geometric size of the slab.
The types of sections that are available to be chosen are user extendable. See
section 5-10 for a description of how to do this.
Page 3 of the quick define box contins the definition for the longitudinal
reinforcement (but not prestressing strands). Bars are selected similarly to the other
prograrns either by area or by name. The bars will be placed into layers if there are too
many to fit within the width of the cross section. Response-2000 uses bar spacing equal
to the bar diarneter to produce layers of steel.
shows the outer ring of the reinforcement and the bottom section shows the inner ring of
reinforcement. In the case shown, the b a n are being defined for a column with
interlocking spirals and the top and bottom rings are being defined.
Tendons are placed in layers as explained above, except that the spacing is
automatically selected as 2 inches (50 mm).
AUXU4.L Page A-1 1
Membrane-2000 Res~onse-2000Triax-2000 Shell-2000
2-2 Defining General Information
Crack spacing in each direction is also defined here. For each direction, the crack
spacing may be selected as either a constant number, or by selecting the check box to
make it automatically calculated. It is recommended that the spacing always be
automaticaIly calculated as it avoids the user fiom having to think about it, and also better
modeIs real behaviocr than a simple constant number.
The equation used for crack spacing at a given depth z is based on the CEB crack
spacing suggestionss and given by the following equation:
where c is diagonal distance to the nearest reinforcement in section fiom current depth
d b is the diameter of the nearest bar
p is the percentage of steel within a depth of z +/- 7.5 db
For cases with no reinforcement, the crack spacing is selected as 5 times the depth
of the section.
iUNUAL Page A-12
Membrane-2000Response-2000 Triax-2000 Shell-2000
Response-2000, as shown, also has an option for the moment axis to be selected.
This represents the depth in the cross section at which any axial load is applied. The
default selection of the centroid of the gross concrete section is generally acceptable, and
if there is no axial load, then this option has no effect.
MANU& Page A-13
Membrane-2000 Response-2000 Triax-2000 Shell-2000
2-3 Materials Definition
Each program defines material properties for three different categories of
materials: concrete, non-prestressed reinforcement and prestressed reinforcement.
Within each category, more than one type may be defined. For example a
prestressed beam in Response-2000 may have 60 MPa concrete for the girder as well as
3 5 MPa concrete for the slab. There may be 1860 MPa low-relaxation steel for the
tendons as well as a 400 MPa steel for the deck reinforcement and 300 MPa steel for the
stirrups. Al1 these material types are defined within the same file.
The first page, as shown here, is the general page. If a material type is fully
defined by default parameters, such as shown here for the concrete fiom panel PV20 in
Membrane-2000, there will be one number showing as the concrete definition. Clicking
on the button to the nght labelled 'metailcd f ;' will allow altering of these default
properties.
AhtAKLAL Page A-14
Membrane-2000 Reswnse-2000 Triax-2000 Shell-2000
If the type has been altered fiom the default values, or if there is more than one
type, then a number will not show up in the general page, d e r , it will list "Detailed" as
above for PV20 reinforcement where there are different steel definitions for the X and Y
directions. To edit the detaled list, click the button beside it. If the detailed title is
replaced with a number, the original list of types will be replaced after a warning
message.
The "predefined type" option allows selection fiom cornmon types of steel
defined in Table 2-2, below, dong with al1 the other parameters used in this dialog box.
Curve is linear to yield, flat post yield, and quadratic after strain hardening.
Slope is zero at location of maximum stress and strain.
Predefined Options
more information 1s
available about the stresostrain properties, however, Ref. 5 provides a method to
calculate the parameters A, B and C as listed in the dialog box.
Predefined Options
Page three of the concrete box in Response-2000 allows any concrete geometry at
al1 to be defined as well as definition of concrete type regions. Sections entered in page
one or two may be "tuned" using page three.
The AddModifjdJ3elete buttons act in the expected way, adding a line to the
listing, modiQing an existing line or removing a line fkom the listing. Note that the
selected line in the listing is shown in red on the sketch on the right side.
To enter a elliptical section, enter the width at the bottom and top extremes, say,
200 mm wide 100 mm up and O mm wide at O mm up as shown on the next page
drawing. Adding in a new line with an elevation between the other two, say 50 mm, and
iWtNUAL Page A-2 1
Membrane-2000 Response-2000 Triax-2000 Shell-2000
a width of " D O W (no quotes) will produce the drawing in the
second figure. Selecting a width of "U'P'' instead will produce
what's shown in the third figure. Using combinations of these,
circles and ellipses may be easily produced. The "up" or "down"
title indicates whether the top or bottom has a zero slope.
Each dialog box uses the traditional list of layers with the ability to add a new
definition, modie an existing one or delete it. This is the same style used in the matenals
definition page.
W A L Page A-22
Membrane-2000 Resnonse-2000 Tnax-2000 Shell-2000
Lndividual Layers
In the example, three layers are defined, with the one called "bot2" cunently
highlighted. It has 3 bars defined each with a cross sectional area of 440 mm2and a
centroid 38 mm above the bottom of the cross section. The type of steel selected is
"botlong" which would have been defined in the matenals dialog page. Different layers
can, of course, use different material types.
Table 2-4 shows the bar types built into the programs. See section 5-8 for a
description of how to add new bar types to this listing.
A4WVUAL Page A-23
Membrane-2000 Response-2000 Tnax-2000 Shell-2000
Table 2-4 Reigorcing Bar and S t r d Designutions
CSA Reinforcing Ban. CSA Pmstressing Strands
"Distributed Layers"
allows a series of
layers to be
automatically
repeated. The
example shows part
of a wall with 15M
bars at 300 mm on
each face. 3 bars per
layer for 6 layers are
used to define this.
Circular patterns,
only available in
Response-2000,
allow reinforcement
to be easily added
for round columns.
This example shows
a large d u m n with
24 # 14 bars at the
listed geometry. The
Orientation specifies the angular offset of the pattern. If the selection is "aligned, then
there will be a bar at the 12-07clockposition on the drawing. If the setting is "offset", as
here, the top 2 bars are balanced around the 12-o'clock position. With more than perhaps
6 bars, this has very little impact, but can be important if there are only, say, 4 bars in the
pattern.
Tendon Lrryers
Response-2000
requires explicit
definition of tendon
layers as contrasteci
to longitudinal
reinforcement layers.
The example shows
a long list of
individual layers for
tendons. Each is
defined as the
number of strands, prestrain, distance fiom bottom of section, type and drape. Drape is
defined as the rise over run of the strands. As such, in 20 feet of run the shown example
would rise 0.0711 x 20 = 1.422 feet rise per 20 feet of run.
Membrane-2000 and Shell-2000 allow prestressed steel fkom the normal layer
dialog box. Draped strands are not supported for shell elernents and membranes. Triax-
2000 reinforcement,
in each direction Y
Y, or 2,is defined
more simply than the
other prograrns as
there is very limited
spacing information
that needs to be
defined. It is simply
defined by the
percentage of steel
- - -
and the bar type.
MINUAL Page A-27
Membrane-2000 Response-2000 Triax-2000 Shel l-2000
2-6 Transverse Reinforcement
Transverse reinforcement, li ke longitudina1 reiriforcement, is defined simi larly
between Shell-2000, Response-2000 and Membrane-2000. Note that Triax-2000 does
not have any definition for transverse reinforcement as in a 3D block of concrete, the
transverse direction is actually the longitudinal Z direction.
Membrane-2000 and Shell-2000 use a similar dialog box with the following
differences. The spacing t e m is replaced by a transverse percentage. The stirmp types
are limited to single-leg hooked bars and single leg t-headed bars. Note that the single-
leg hooked bars are currently drawn on the screen as t-heads.
M A L Page A-28
Membrane-2000 Reswnse-2000 Tnax-2000 Shell-2000
2-7 Element Catalog
While Response-2000 only allows one cross section per input file, the other
programs al1 allow more than one by use of the cataiog menu option.
The example
here shows the catalog
in use with Membrane-
2000 showing a Iist of
experimental tests.
Shell element SE5 is
currently selected from
the listing. The catalog
is based on the familiar
Windows Explorer tree-
systern. The different
titles used are fiom the
"Edit General" page in
the define menu.
The catalog buttons on the right allow a new element to be created using the
Quick Define Wizard, copying of an existing element, or deleting of an element from the
catalog.
When using the programs, it is possible to switch to a different element via either
the catalog itself, the menu options "Catalog ] Next Element", "Catalog 1 Previous
Element", or using the toolbar. This fiagment of the toolbar, shown here fiom
Membrane-2000 allows access to the catdog fiom the button that looks like a little tree-
Iist between the arrows. The arrow pointing lefi goes to the previous element in the
listing, and the m o w to
the right goes to the next
element in the lia. In this way it is easy to examine many elements from within one file.
h4HMXA Page A-29
Membrane-2000 Response-2000 Triax-2000 Shell-2000
3-1 Membrane-2000
Lding
Loading for Membrane-2000 consists
Axial stress in the X direction, Axial
stress in the Y direction, and in-plane
shear. Positive axial stresses indicate
tension with negative indicating
compression. The shear must be non-zero
and positive in Membrane-2000.
merimental Remlis
Because Membrane-2000 represents the type of element tested to define the
MCFT, a facility has been included to allow experirnental results to be shown. This is
demonstrated in Section I, Quick Start for Membrane-2000 where the results fkom panel
PV20 tested by Vecchio are stiawn.
The expenmental results are added one variable at a time- A didog box allows
access to 12 variables. Expenmental data in the fonn of a column of numbers may be
entered by hand or using the "paste" button on the page. There must be the same nurnber
of data points for each variable and they must be in the same order. When an analysis is
mn,Membrane-2000 checks if data has been entered for both the X and Y axes of the
plots. If so, it includes the experimental data dong with the caiculated solution.
The last menu option in the loads menu of Membrane-2000 allows the data to be
quickly removed fiom, for example, the defauh input example.
hding
As an analysis program of a
general 3D solid, Tnax-2000
requires 6 loads to be defined
to perform an analysis. These
are Axial force in the X, Y, and
Z direction plus shear on the X-
Y, Y-Z,and X-Zplanes.
As with the other programs, the first colurnn is for the initial loading or single
load level analysis. The second column is used for ratios h e e n the loads for a fiil1
response trpe o f anaiysis.
Loading
Shell-2000 allows al1 8 force resultants on a shell element to be applied. The
loads are applied in force resultant per unit length. For example, moments are in kNm/m
or kip-ft/ft and axial forces are in kN/m or kipslft. The applied loads are:
Component Sign
. . - .-- -
For this example, there is no initial load level, and the moment to shear ratio is
1.34 feet.
This module will ody be u d if the checkbox at the top in selected. The age for
long-tenn behaviour is needed, as is the sustained moment for the section as that strongly
affects the creep.
Briefly, the shrinkage and relaxation is estimated for the given age and the creep
under the given sustained moment is estimated. Then a shrinkage/thermal profile is
automatically added to the section to mode1 this. Analyses done then represent shon t e m
loading on a well-aged beam or colurnn. For a more detailed description of the time-
dependent effects module, see Reference 2.
The top right side allows setting of individual thermal strains for the
reinforcement. As shown, it is also possible to select a value and apply it to al1 layers of
AhAWAl: Page A-3 5
Membrane-2000 Response-2000 Triax-2000 Shell-2000
reinforcement. Note that these mains can be used the same way as prestrains are to
tendons if desired.
The bottom nght shows a plot with a line indicating the shrinkage distribution and
with little dots to indicate the thermal strains of the reinforcement.
The example shows a 120 inch high section with a large distribution of thermal
strains in the top as well as a small distribution in the bottom. The reinforcernent does
not have any thermal mains defined for it.
iWWUAL Page A-36
Membrane-2000 Res~onse-2000Triax-2000 Shell-2000
Strain Discontimriy
The strain dircontinuity dialog box allows modelling of behaviour effects due to
composite construction. See section 5-7 for a description and a a m p l e of how best to use
the strain discontinuity feature. In g e n e d , it allows for an explicit difference between
the longitudinal strain profile at a given depth and the basic assumption that plane
sections remain plane.
The interface is similar to the shrinkage page above. Elevation-main pairs are
added to the list and they are plotted on the graph at the right.
The right side has similar options. In this case, the fixed support also needs
information about the penetration of strains into the bottom block of concrete that would
be supporting the wlumn. The default value of 0.022 is suggested for columns. See
Reference 2 for an explanation of how this is used.
M A L Page A-38
Membrane-2000 Response-2000 Triax-2000 Shell-2000
Membrane-2000 and Triax-2OO perform their analysis immediately and let the
user observe the results. Response-2000 and Shell-2000, due to longer execution times,
update the screen as new results are obtained and let the user watch the results as they
arrive. The different screens of data presentation are available from the "view" menu in
the prograrns or, more wnveniently, through the toolbar.
The default data presentation mode in the programs is a screen with 9 graphs on
it. This manual will concentrate on that "9-plot" view. The program also has a mode to
allow the same plots to be shown individually. For Triax-2000 and Membrane-2000, the
plots in the 9-plot mode are generally full load-deformation plots. For Response-2000
and Shell-2000, the plots represent one load stage showing the value of the variable over
the depth of the element. An additional page is available that provide load-deformation
plots for the latter programs.
AUNUAL Page A-39
Membrane-2000 Response-2000 Triax-2000 Shell-2000
The large load-deformation plots have an additional feature in that they allow
other plots to be pasted ont0 them. The example shown was generate with Shell-2000 in
a minute or so. It shows a shell element subjected to in-plane shear and moment. The
initial cracks are fleniral in nature and as the cracks reach a larger depth, the slope of the
moment-curvature plots changes dmatically. The cracks can be seen in the final form
with yielding steel in the 1 s t little plot.
The interface of the programs is set up so that there is a "control plot" on the
lower left of the screen. (2 control plots in Response-2000). These are used to indicate
whicti pan of the analysis is currently being viewed. To change the currently viewed
load, click on the control plot and move the crosshairs. Pressing Page-Up and Page-
Down will also switch between differently viewed load nages. The button "madauto
range" will automatically adjust the plots to "remember" the maximum s a l e over the
course of the analysis for easier scaling.
iM4NUAL Page A 4 0
Membrane-2000 Response-2000 Triax-2000 Shell-2000
4-2 Types of Analyses
Three types of analysis are cornmon to al1 the program. The first is the "Full-
Response" type of analysis. This will first do a single-load Ievel analysis at the values in
the left colurnn of the '2oads 1 Loads" menu choice, and then increase loads in the ratio
shown in the right column of the '2oads 1 Loads" menu item.
The second type of analysis is a "Single Load Level". This will solve to the loads
selected in the lefi side of the "Loads 1 Loads" dialog box.
The third type of analysis is a strain state analysis that will return the stress and
force state that corresponds to a given set of global strains.
Membrane-2000 also includes explicit options in the solve menu to peflonn a full
analysis for a number of analysis types. These include the Modified Compression Field
Theory 1987 (MCFT)', the Rotating-Angle Softened Tmss Model m - S T M ) 1993",
,the Fixed Angle Softened Tmss Model (FA-
the RA-STM 199~'~
the RA-STM 199~'~,
, F A - S T ' 1 9 9 7 ~and
S T M ) 1996~' ~ FA-STM 1998~.
Hsu and colleagues at the
University of Houston derived the last 6 methods. They are included in Membrane-2000
for cornparison purposes. In general, the six methods nom Houston do no better a job
than the MCFT,despite having much more experimental data to derive fiom.
-
Shear y, In-plane shear stress @Pa, p s i kg/cm2) vs Shear strain (x IO'-')
As Membrane-2000 uses shear for al1 analyses, this summarises the element
behaviour. Note cracking, tension stiffening, and crushing at 4.5 MPa in the example
hWVU.AL Page A 4 2
Membrane-2000 Response-2000 Tnax-2000 Shell-2000
Shear f ,- In-plane shear stress vs X direction Average steel stress (MPa, ksi, t/cm2)
The average steel stress may not exceed the yield stress except due t o strain
hardening. Note the unloading and the change in dope when the y direction steel yields
Crack Diagram Plot of crack patterns with crack widths (mm, in, cm)
Drawn diagram is 5 x panel thickness wide with steel and cracks drawn in to
scale. Note that cracks rotate with analysis, as per MCFT.
Shear f, - In-plane shear stress vs Y direction Average Steel stress (MPa, ksi, t/cm2)
Same as the above. Note that the steel does not exceed yield (297 MPa).
-
Shear 0 Shear stress (MF* psi, kg/cm2) vs Principal angle of stresdstrain (deg)
This shows the angle rotation during the analysis.
Node-Data Stress and strain state of panel at given load stage (units shown)
By right clicking, it is possible to copy this data to the Windows clipboard for use
in other applications.
Shear - ~ 2 Shear
, - el
Similar to graphs on first page of 9 plots, but with shear as vertical axis.
MWUAL Page A-46
Membrane-2000 Resoonse-2000 Triax-2000 SheN-2000
Membrane-2000: 9 Plots Rdar and Cracks
This final page of 9 plots fiom Membrane-2000 shows important details of the
MCFT in the average steel stress and steel stress at a crack plots.
The important thing to note about this page is the "stress at a crack" plots. The
crosshairs in the control chart have been selected here to be just before shear on the crack
is required. This can be seen on the centre right plot where the cross is just before yield
in the weak (Y) direction. This is an assumption of the MCFT;no shear on the crack is
needed until steel first yields at a crack in the weaker direction. Put another way, the
shear on the crack is minimisai in the MCFT as it is assumed that the reinforcement
mechanism is stiffer.
As the load increases above that point, the stress at a crack in the Y direction is
limited to the yield value, and the average is allowed to continue to increase towards
yield. Together this means that the amount of shear on the crack must increase to
continue to support the principal tension. Note that as the load increases above the
A-4iNUALPage A-47
Membrane-2000 Response-2000 Tnax-2000 Shell-2000
current level, the rate of change of the stress at a crack in the X direction changes. Shear
on the crack acts to transfer reinforcement demand at a crack fiom the weak direction (Y)
to the strong direction (X), so shear on the crack makes the dope change. If the load
could continue to increase higher, in this case it's govemed by fnuhing of the concrete,
and the steel stress in the strong direction also reached yield, the principal tensile stress
would be Iowered to ensure equilibrium couid be maintained locally at a crack. That, in a
nutshell, is the point of the crack check in the MCFT.
MANUAL Page A 4 8
Membrane-2000 Response-2000 Triax-2000 Shell-2000
Response-2000 uses two control plots. They are selected based on the type of
loading, but for shear analyses, the top one shows shear versus shear-strain plot and the
bottom one shows the moment curvature plot. This quickly allows detection of shear
failures versus flexural failures. In this case, due to the prestressing, the moment
curvature never actually reaches a positive curvature, but the shear plot has started
descending for increasing strain. This indicates a shear failure in this case before even
fb11 depth cracking of the section. The plots show the behaviour just before failure.
Cross Section
M A L Page A-49
Membrane-2000 Response-2000 Triax-2000 Shell-2000
The cross section is drawn darker in regions where the concrete has not cracked.
Longitudinal reinforcement and stirrups are draw dark red if on the yield plateau, bright
red if strain hardening, and dark and bright green for yielding in compression. In this
case, note that despite the positive moment on the section, the bottom of the section has
not cracked through yet due to the prestress force.
Crack Diagram
This plot shows the estimated crack pattern as well as crack widths (mm, in, cm).
Note that the crack widths as well a s patterns are rather approximate and should no: 5e
used alone t o estimate the health o f a structure. For this beam about to fail, the maximum
crack width is predicted to be 3.O mm. For cases where part of the concrete is crushing,
the section is redrawn in pink, and for sections where the cracks are slipping causing
failure, the section is drawn in purple.
Response-2000 calculates crack spacing based on the angle and the estimate of
crack spacing in the longitudinal and transverse directions as per the MCFT. If the crack
spacing is calculated automatically as suggested, the spacing will Vary over the depth of
the section, hnher improving the realism of the analysis.
This page has 2 new graphs that have not been explained yet.
The Interna1 Forces plot shows the force and location of the compressive and
tensile forces in the cross section. In this case, due to the axial load, they do not balance
each other. Note that the tensile force arrow may not corne directly fiom the steel
location due to the concrete tensile force component. By right clicking on the plot,
another mode may be selected that calculates directly the resultant of the steel and
concrete forces. This can produce counterintuitive results, so is not the default mode of
presentation.
The N+M plot shows the moment and axial force drawn simply as arrows. This
helps in finding mistakes in simple things such as the sign of the axial force.
ibWVUAt Page A 4 4
Membrane-2000 Response-2000 Triax-2000 Shell-2000
The background plot is selected in the top left of the screen. Cumently it is
selected to the AASHTO-99 LRFD M-V interaction diagram. Below that option is the
"Paste Data" section that allows selection of which plots to paste ont0 the plot. The
figure at the start of the section showing cracked shell elements on the main figure was
A4MNU.a Page A-55
Membrane-2000 Response-2000 Triax-2000 Shell-2000
prepared this way with Shell-2000. Currently selected is the longitudinal strain profile,
with the current levels controlled by the control plot below. Pressing the "add to graph"
button would paste the picture on the main figure where it could be moved and resized.
The main figure wntains a pair of text-boxes as well as a diagram of the element.
The element picture is pasted on via the "Options / Insert Beam Diagram" menu option.
By right clicking on the figure, the dimension text may be resized and copied to the
clipboard etc. The top text was automatically prepared by Response-2000 and inserted
with the "Options 1 Insert Text Box'' option. This text box may also be edited and
customised. The bottom right text box was automatically included by Response-2000 to
provide information on how the AASHTO-99 analysis was calculated.
There is an additional littie box shown just below the envelope, this is the
maximum load that the Response-2000 analysis was able to achieve. In this case they are
very close, but they can Vary more widely. If the Response-2000 prediction is outside the
envelope, it suggests that the AASHTO code is conservative compared to the more
advanced predictions that Response-2000 makes. If the Response-2000 prediction is
within the envelope, it suggests that the code is unconservative compared to the
Response-2000 analysis. This provides a second, independent, checking of the
provisions of the code that uui add to engineer's confidence for strength predictions of
unusual geornetry.
W A l : Page A-56
Membrane-2000 R c s p o n ~ 2 0 0 0Triax-2000 SheH-2000
It is noted in the top text box that this particular column happened to fail
experimentally at a shear of 80 kips, which is in excellent agreement with both the
Response-2000 predictions and the code prediction. See Reference 2 for more discussion
of the experimental veri fication of Response-2000.
To perform the shown Response-2000 analysis, the section was first entered into
the program- Next, the "Loads 1 Full Member Properties" option was used to select a
length of 18 metres (Response-2000 does the analysis on the half-length of the beam),
and the loading was switched to a uniformly distributed load. Finally, the "Solve 1
Member Response" menu option was selected.
Response-2000 calculated the interaction diagram shown in the top control chart.
It then detemined the largest loading envelope that would fit into the diagram. It can be
Ad4NUAL Page A-58
Membrane-2000 Response-2000 Tnax-2000 Shell-2000
seen that the loading envelope touches the filure envelope on the top indicating a shear
failure. If it had touched at the right side, it would have represented a flexural failure.
The shape of the loading envelope is parabolic on the right and Iinear on the left. See
Reference 2 for a description of the derivation of the loading shape.
The plots at the bottom are also instructive. The top left one shows the change in
curvature over the length of the beam. The location of first flewal cracking, about 2500
mm from the support, c m be clearly seen as can the roughiy parabolic distribution that
would be expected for the parabolic moment diagram. Note that these curvatures al1
implicitly include the effect of shear on the cuwature.
The shear strain distribution shows ihat the average shear strain over the length of
the beam is not uniform at all. It may be expected that the strain would increase linearly
fiom the right as the shear diagram is linear, but this is not the case due the concrete non-
linearity. The strong interaction of shear and moment for this beam means that the
predicted cntical location for shear is about 6 metres away fiom the location of maximum
shear.
Rounding out the plots are the predicted deflected shape and the plot of load
versus deflection for the beam. Note that the failure is predicted at a load of 102.6 kN/m
and a deflection of about 100 millirnetres. The expermental failure load was measured
as 105 kN/m, at a deflection of about 100 millimetres.
iW.NUfiPage A-59
Membrane-2000 Response-2000 Triax-2000 Shell-2000
The example shows a block of 40 MPa concrete with 1 % of 400 MPa steel in the
X direction, and 0.5% of 400 MPa reinforcement in the Y and Z directions. The loading
is a full response with 4 units of shear stress in the X-Y direction and 5 units in the Y-Z
direction.
Crack Diagram
This shows the block of concrete with the reinforcement drawn in. The
reinforcement is given the same colouring d e s that Response-2000 has. The directions
of the vectors of principal stress/strain are drawn in as weli, thick blue for the principal
ibU.NUALPage A 4 0
Membrane-2000 Response-2000 Triax-2000 Shell-2000
tensile direction, thick green for principal compressive direction and red for the
intermediate direction. If the concrete is cracked, the intersection of the m c k plane with
the bounds of the box is s h o w with colours represmting the principal direction, in this
case blue due to it being the first principal tensile direction.
Node-Data 1 and 2
Like Membrane-2000, Triax-2000 provides a page that shows the interna1 stress
and strain state of the element. For the 3D case, this presentation is sufficiently imporiant
to be on almost al1 of the 9-plot cases. With a right click on each of the 2 pages of
numbers, the data may be copied to the clipboard for use in other programs.
Load Factor- Crack width 1 load factor VS crack width I (mm, in, cm)
Crack widths are s h o w here for the first principal direction of cracking. Note
that there rnay well be more cracks in the second principal direction and, in the absence
of shear, on the third direction also.
Cracking: This %plot view shows the principal strains (x IO"'), crack widths (mm, in,
cm) and shear on crack resultants (MPa, psi, kg/cm2), for cracks in principal directions
one and two.
Direction Cosines This shows the state of the three direction cosines in each of the
three principal directions with respect to load factor. These are the numbers used to draw
the vectors in the crack diagram above.
Rein forcement This group of 9-plots shows the steel stresses on average and at a
crack (MPa, ksi, t/cm2)in the X, Y and Z directions. Note that the discussion above for
Membrane-2000 conceming shear on the crack also applies to effects in three dimensions
such as these cases.
MihWRL Page A-62
Membrane-2000 Response-2000 Tnax-2000 Shell-2000
Shear on the Crack This page shows the 3D components of the shear on the crack
(MPa, psi, kg/cm2)for the first and second principal cracking direction. These values are
calculated using a non-hear optimisation fnction based on quadratic programming. It
minimises to the shear on the crack required, an extension of the 2D method explained
above for Membrane-2000.
Strain State This shows the load-factor vs the 6 rectangular strains. ( E , E, cl, yw, yYr,
, ,y al1 10-~)
Stress Statt This shows the load-factor vs the 6 rectangular stresses. (Ex, &, &, vxy,
,,v v , al1 in MPa, psi, kg/cm2)
Sheii-2000 9 plois Genetaf
Like Tnax-2000, the 3D nature of Shell-2000 re~ultsin eight 9-plot m e n s .
Shell-2000 is shown with Shell element SP2 tested by ~ d e b . 2with
~ loading ratios of
moment M x = My = 0.5 and out of plane shear Va= V , = 1 .O. This element containecl a
light amount (53 psi = 0.37 MPa) of transverse reinforcement.
It can be seem that the combination ofthe two moments and shears has causeci the
direction of cracking to rotate to 45 degrees away fiom the steel directions. The applied
loading at predicted failure is V
, =V
, = 303 kN/m. In this case, this corresponds to a
principal shear o f 3O3 x 1 -414 = 428 kN/m. In the expenment, the element failed at a
principal shear o f 449 kN/m.
lhXNiXAL Page A-64
Membrane-2000 Response-2000 Triax-2000 Shell-2000
Shell Diagram
The crack diagram shows the extent of cracking in the shell element. Bars are
drawn and turn red or green when yielding as in the other programs in this manual. The
fust principal direction crack is drawn in red and the second principal direction crack is
drawn in green. The Crack diagram is s h o w below responding t o each of the different
type of loading to assist in interpreting what it means.
Crack Width 1 width of crack in first principal direction (mm, in, cm)
This shows the crack width distribution over the element
Shear Stress V
, In plane shear distribution with depth (MPa, psi kg/cm2)
Shell elements can be subject to intemal in-plane shear stresses even when not
externally loaded in in-plane shear. The example shows that in this case there is up to 3.5
MPa of in-plane shear even though the resultant ofthe positive and negative in-plane
shear is zero. The resultant o f the twisting moment causeci by this shear is also zero as
t here is no extemal twisting moment applied.
Direction Cosines This shows the state of the three direction cosines in each of the
three principal directions with respect to depth. These are the numbers used to draw the
crack surface in the shell diagram.
W R L Page A 4 6
Membrane-2000 Response-2000 Triax-2000 Shell-2000
Principal This page shows the principal strains (x 105) and principal stresses @Pa,
psi, kg/cm2) with respect to depth.
Reinforcemcnt This shows the steel stresses on average and at a crack as well as
bond stresses va, ksi, tkm3 in the X, Y and Z directions. The bond is discussed
above in the Response-2000 section on bond. The discussion above for Membrane-2000
conceming shear on the crack also applies to effects in three dimensions such as these
cases.
Shear on the Crack This page shows the 3D components of the shear on the crack
@Pa, psi, kg/cmz) for the first and second principal cracking direction. These values are
calculated using a non-linear optimisation hinction based on quaciratic programming. It
minimises to the shear on the crack required.
Strain State This shows the 6 rectangular strains through the depth o f the element.
(EX? Gy, Ez, Y?
, Y,=, Y a . x W3)
Stress State This shows the 6 rectangular stresses through the depth of the element.
(f& f, fa,,,v v, ,v al1 in M'Pa, psi, kg/cm2)
M4AUAL Page A-67
Membrane-2000 Response-2000 Triax-2000 Shell-2000
Combinations of loading:
Axial load in X and Y direction
Two crack planes form at 90 degrees.
For example, the following series of letters in a text box "a-bcAdjelf' would be
shown on the screen as:
abcda
iv) Properties
This allows editing of the limits, fonts, titles etc of a chart. See section 5-3 below.
v) Auto* Range
This will change the axis mode of the chart. Normally charts automatically re-
scale the axes so that the data f i t ~nicely on the chart. When switched to max scaling, the
chart will "remembef' the maximum scale values so far. This is usefbl when scanning
through al1 the results with the control plot. This hnction is duplicated with the "auto
range" button on the lefi side of the screen above the control plots that works on al1 charts
sirnultaneously.
Title Section
The chart title, X-axis title
and Y-axis title rnay be changed
here. Note that the methods explained in section 5-11 ifor supedsubscripts and Greek
characters rnay be used (which rnay explain why the text rnay look strange for some
charts). The small charts used in the 9-plot output do not show their axis titles, but they
are stored here anyway so that units and axes rnay be confrmed. The user can change the
fonts for the title and axes fiom this part of the dialog box as well.
To account for the slab, extrapolate the Iinear strain profile into the location of
where the slab will be. If the slab were to be 8 inches deep, placed directly on top o f the
beam, the strain in the bottom of the slab would equal the above value of -0.1593 m d m ,
and the strain at the top of the slab would equal -0.1273 mm/m.
AU.NUAL Page A-74
Membrane-2000 Response-2000 Triax-2000 Shell-2000
Now enter in the slab itself, 80 inches wide and 8 inches thick. For this example,
The strain graph shows that while plane sections are still plane, there is an offset
that happens to rnake the strain equal to zero in the top slab section up to the top of the
axis. The effect of this on the stress plot is that there is no stress in the slab- This means
the stress in the beam is still the same as it was before the addition of the slab, and there
is no stress in the slab itself
-0.22 844.3
bat,
A&lNUU Page A-76
Membrane-2000 Response-2000 Triax-2000 Shell-2000
5-8 Rebar.dat
Each of the programs has a list of reinforcing bar definitions that can be used by
any cross section (see Table 2-4 in section II). This list is, in fact, user definable. Each
program maintains a file in its install directory called "rebardat" that is a text file loaded
each time the program starts. Users may add to this lia and the new options will be
available the next time the prograrn is started. Note that each prograrn has a separate
rebar-dat, but they are ail identical on distribution. (Le. if changes are made to one, they
can be copied to the directories of the other programs as well.)
/ / Response-2000 Data F i l e
//
// T h i s f i l e c o n t a i n s the d e f i n i t i o n s of a i l s t a n d a r d r e b a r / s t r a n d types
//
// U s e r s may add more t y p e s which w i l l be a v a i l a b l e t h e n e x t t f m e c h a t
// Response-2000 is s t a r t e d . i n p u t is n o t c a s e s e n s i t i v e .
//
// I f bars a r e e n t e r e d w i t h t h e same namc a s e x i s t i n g o n e s , t h e first o n e v i l 1 be used
// Lar t i t l e is l i m i t e d t o 1 4 c h a r a c t e r s . S p a c e s a r e a l l o w e d , b u t t h e f i r s t nuniber
// found a f t e r t h e t i t l e a n d a s p a c e i s assumed t o b e t h e a r e a .
//
// I n f o r m a t i o n is a s f o l l o w s :
//
// Marne code Nominal Diameter (mm) Nominal Area (mm-2)
//
// -- s t a r t o f d e f a u l t l i s t i n g --
//
// CSA s t a n d a r d R e i n f o r c i n q B a r s
//
10M 11.3 100
15M 16.0 200
ZOM 19.5 300
25M 25.2 500
30M 29.9 700
35M 35.7 1000
-. e t c
If a file including a user defined bar is used on a version of one of the programs
that has not seen the bar title before, the new name will be saved to the standard listing
when the program shuts down.
Iic.IIANuAL Page A-77
Membrane-2000 Response-2000 Triax-2000 Shell-2000
5-9 Adding predefined shapes: Sbapedat
The list of basic shapes used in the concrete definition in Response-2000 is user
extendable. This may be usefl for things such as design optimisation where the same
geometry may be tweaked by adjusting only one or twr> variables.
// This file defines al1 the simple shapes used by Response-2000
// note that the standard shapes (PCI etc) are located in standard-dat
//
// L'sers can add more shapes to end of this file. They will be available
// the next time the proqram is loaded
//
// The format here is deffned as follows:
//
// TYPE name title of section
// line one text line describinq the section
// name-ico (optional) name of icon f rom current directory to use
// ( i f not included a default user icon will be used)
// P A W S b h variable names to define the shape
// lines one line for each variable as text definition
// SECTION start of the section definition
// height width one line for each change in geometry
// simple math is allowed ( +- '/ 1
// REBAR RECT rebar pattern symbols allowed: RECT, POLAR, BIPOLAR
// RECT means rectanqular reinforcment (like a bearn)
// FOLAR means circulas (like in a tound colurnn)
// BIPOLAR means interlockinq spirals pattern
// ENDTYPF indicatinq end of the list
//
// half circles and elliptical parts can be defined as:
//
// heiqht width
// code
// height width
// with code = U P or GOWN indicatinq uhich side has the zero slope
//
// Absolute dimensions in m. Distance from bottom of section
// lines startinq with 2 slashes are comment lines
//
TYPE RECT
Rectangle Section
PhRAMS b h
Width of the Rectangle
i-ieiqht of the Rectanqle
SECTION SOLID
h b
O b
SEBAR RECT
ENDTY PE
/ /
TYPE INTER
Interlockinq Spiral Column
inter. ico
PARA% b h bt
Total section width
Total section heiqht
Uidth at top and bottom extremes
SECTION SOLID
h bt
h-(b-bt)'0.5 b
(b-bt)'0.5 b
O bt
REBAR BIPOLAR
ENDTY PE
AhdNiYAL Page A-78
Membrane-2000 Response-2000 Triax-2000 Shell-2000
5-10 Adding predefined sections: Standard.dat
This file is very similar to the shape-dat file explained above except that it does
not allow variables to define the section. It is used ta define standard sections such as the
AASHTO highway girder cross sections.
Users can add more to end of file as well with the following format
heiqht width
code
heiqht width
with code = U P o r DOW indicating uhich side has the zero slope
TYPE CPC1900
CATEGORY C P C I - 1
SECTION SOLID
?O0 300
750 30C
720 150
240 150
150 450
O 450
ENDTY PE
//
M L Page A-79
Membrane-2000 Response-2000 Triax-2000 Shell-2000
5-11Template Files
When the programs are started, they look automatically for a template file, to load
up default values for the program to run with.
The template file mua be located in the directory ofthe program itself. The
following parameters are set tiom the template file:
Default units
Default name for the "done by" part of input files
Default steel yield stress
Default aggregate size
Default concrete strength
To create a template file, simply make an input file that has al1 the desired base
components generally wanted in a starting file. For some, this may include matenal
types, concrete geometry, loading, units, etc. Save this file in the program install
directory with the name listed above. Upon restarting the program, it should
automatically load this new file. Note that Response-2000 will automatically rename the
file "r2k.r2k.rspmas .rsp is the standard file extension for Response-2000. This means
that the Response template file must be renamed by hand using, for example, the
windows explorer.
This is the simplest way to have the program begin each time with, say, Japanese
units rather than SI metnc, which is the default starting units.
MUVUAL Page A-80
Response-2000 also supports a text-based file format that supports al1 the features
of the hl1 binary version of the program. What foilows is a definition of this format with
an example as well.
Usage
This file format can be transparently used in Response-2000. The existing load
dialog box in Response-2000 allows access via the "files of type" option.
Extension
Filename extension is ".r2tn(Response-2000 text file).
Contents:
The following is a comprehensive list of the contents of the file. Note that options
within square brackets ([ j) are optional and Response-2000 will assumed default
numbers for them if not provided. Lines preceded by a "#" or "//" are comment lines and
may be inserted at d l .
Version:
The format described herein represents Response file format number 0.8. This
version type will remain supported for al1 released versions of Response-2000.
Units
Units for input are listed below in order of Metnc, US Customary, and Old Metnc
(ex mm,in,cm means millimetres in SI metric, inches in US Customary, and centimetres
for kglcm2 units)
ENDMAT
[... a s many a s 5 c o n c r e t e m a t e r i a l t y p e s i n t o t a l r e p e a t i n g t y p e l i s t a b o v e ]
MATERIAL REBAR - t e x t l f i r s t r e b a r d e f i n i t i o n b l o c k w i t h name
EY -fnum yield stress (HPa, k s i , t o n s / c m 2 )
[E -fnuml Young8s modulus iMPa.ksi.tons/un21
[FU -fnum] u l t i m a t e s l r e n g t h (MPa, k s i , t o n s / c m 2 )
[ESH -fnuml s t r a i n a t s t a r t o f s t r a i n h a r d e n i x 10-3)
[EU -inun11 s t r a i n a t p e a k s t r e s s ( x 10-3)
EN DMAT
MATERIAL PRESTRESS - t e x c l f i r s t p r e s t r e s s e d steel t y p e d e f i n i t i o n
N -fnum u l t i m a t e s t r e n g t h (MPa, k s i , t o n s / c m 2
[E -fnurn] Young* s modulus [MPa, k s i , t o n s / c m 2 )
[ A -numl Ramberg-Osgood p a r a m e t e r A
[ B -numl Ramberg-Osgood p a r a m e t e r B
[C fnum] Ramberg-Osgood p a r a m e t e r C
EU--^
num] s t r a i n a t p e a k s t r e s s ( x 10-3)
ENDMAT
[.. a s many a s a t o t a l o f 20 d i f f e r e n t r e b a r and p r e s t r e s s t y p e s a l l o w e d ]
SECTION SOLI D define t h e solid cross section
-
fnum -fnum - t e x t l d e p t h . u i d t h , t y p e name (mm, i n , c m )
-fnum -fnum - t e x t l d e p t h , w i d t h , t y p e name
[.. . a s rnany a s 30 d i f f e r e n t d e f i n i t i o n s a s s u c h ]
[... a l t e r n a t e m e t h o d o l o g y below. .. use one o r o t h e r j
SECTNAME - t e x t l s h a p e name f r o m s h a p e s - d a t
PARAMS -fnum -fnum -fnurn ... v a r a i b l e p a r a m e t e r s f o r t h a t
ENDSECTION
SECTION HOLLOW
u n s u p p o r t e d f o r now
ENDSECTION
LONGTA3 longitudinal s t e e l table
fnum -fnurn -fnum - t e x t l height, area, prestrain, type of s t e e l
[.., a s many l a y e r s a s d e s i r e d . c a n b e u s e d w i t h LONGREINF o r w i t h o u t ]
LONGREINF - t e x t l f i r s t type of longitudinal reinforcment
Z -fnum d e p t h rom b o t t o m o f s e c t i o n (mm, i n , c m )
A -fnurn a r e a o f s c e e l a t d e p t h (mmZ,in2,cm2]
TYPE - t e x t l type ( e i t h e r rebar o r prestressed type)
[ DRAPE -fnum] drape o f reinforcrnent [ rise over r u n )
[num -fnuml number o f i n d i v i d u a l b a r s
[ A I -fnum] area of individual bar
[DE -fnum] diameter of individual b a r
[BART - t e x t l ] t i t l e of b a r type
Note t h a t we c a n s p e c i f y num=10, b a r t = 1 5
o r we c o u l d s p e c i f y num=lO, AI=200rnm2, DB=16 mm
and g e t same t h i n g . I f b o t h s p e c i f i e d , BART p r e c e d e n c e
[ DEP - f n u n ] Delta-epsilon-p p r e s t a i n (x10-3)
The f o i l o w i n g a l l o w s g r o u p i n g o f b a r s . A l 1 l a y e r s m u s t b e
i n d i v i d u a l l y e n t e r r e d , b u t t h e y c a n be g r o u p e d i n t o
distributed patterns o r circular patterns.
[ PATTERN -i num 1 ( O = none, 1= c i r c u l a r , Z = d i s t r i b u t e d )
[NROUND -1numI number i n p a t t e r n
f o r p a t t e r n x l , i t ' s number a r o u n d c i r c l e
f o r p a t t e r n = Z , i t ' s t o t a l number of b a r s i n i a y e r s
[ALIGNED -inum] i f 1, t h e n 2 b a r s a t t o p , i f O, 1 b a r a t
top for circular patterns
[INDEX -inum] which p a t t e r n number we a r e d e a l i n g w i t h
ENDLONG
[.. a s many l o n g i t u d i n a l p a t t e r n s a s w i l l f i t i n t o mernory a l l o w e d ]
ih4MLM Page A-82
Membrane-2000 Response-2000 Tnax-2000 Shell-2000
TWSREINF - t e x t l f i r s t transverse reinforcement type
A -fnum t o t a l a r e a of a l 1 l e g s of s t e e l
TYPE - t e x t l type ( e i t h e r r e b a r o r p r e s t r e s s e d t y p e )
PATTERN -inum s t y l e of type:
O = single leg
1 = open s t i r r u p
2 = closed s r i r r u p
3 = hoop
4 = T headed s i n g l e l e g
SPACE -fnum s p a c i n g of r e i n f o r c e m e n t (mm, i n , cm)
[DISTTOF -fnum] d i s t a n c e ftom b o t t o m o f s e c t i o n t o t o p
o f r e i n f o r c e m e n t (mm, i n , cm1
[ DISTBOT -f numl d i s t a n c e from b o t t o m o f s e c t i o n t o
bottom o f r e i n f o r c e m e n t (mm,i n , c m )
[ A I -fnuml a r e a of i n d i v i d u a l b a r
[DB -fnum] diameter o f i n d i v i d u a l bar
[SART - t e x t l l t i t l e of b a r type
Note t h a t we c a n s p e c i f y b a r t t l 5
o r we c o u l d s p e c i f y AI=200mm2, DB=16 mm
and qet same t h i n g . I f both s p e c i f i e d , B M T precedence
[DEP -fnum] D e l t a - e p s i l o n - p p r e s t a i n (%IO-3)
ENDTRANS
LOADING
[CONSTANT -fnum -fnum -fnuml c o n s t a n t l o a d component
VARIABLE -fnum -fnum -fnum v a r i a b l e l o a d component
ENDLOAD
[SHRINKTHERM CONCRETE]
[-fnum -fnuml d e p t h and s h r i n k a g e amount
[mm, i n , c m ) , ( x 10-3)
f . . a s many l a y e r s o f s h r i n k a g e a s d e s i r e d f o r c o n c r e t e ]
IENDSHRINK]
[THE.SMAL REINFORCE]
(not supported yet 1
[-fnum] t h e r m a l s t r a i n f o r e a c h l o n g . cype
[ ENDSHRINK]
[ DISCONT]
[-fnum -f nurn] d e p t h and d i s c o n t i n u i t y s t r a i n
[ .. a s many a s u a n t e d ]
[ENDDISCONTJ
WALYSIS
TYPECODE -inurn -fnum -fnum type of a n a l y s i s t o perform
and i n p u t v a l u e s a s l i s t e d
O = none
1 = f u l l response
2 = one l o a d
3 = 2 strain solution
(params s t r a i n , d e p t h , s t r a i n d e p t h l
4 * I strain solution
(params s t r a i n d e p t h )
5 = M-N I n t e r a c t i o n
6 = M-V I n t e r a c t i o n
7 = N-V I n t e r a c t i o n
8 = Pushover a n a l y s i s
[. .a s many a n a l y s e s a s d e s i r e d f o r t h i s s e c t i o n ]
i o n l y used f o r command l i n e v e r s i o n o f r e s p o n s e . Send e m a i l ]
ENDANALYS I S
IMEMBERINFO]
[L -fnum] l e n g t h o f s h e a r s p a n (mm, i n , c m )
[MID002 -fnurn] middle l e n g t h o f c o n s t a n t moment
mm, i n , cm]
[TYPE -inum] t y p e of l o a d i n g
l= c o n s t a n t s h e a r a n a l y s i s
2= UDL beam t y p e
3= UDL f o o t i n g t y p e
[LEETPERCENT -fnum] l e f t s i d e % moment o f r i g h t s i d e ( 8 )
[LEFT -inum] l e f t s i d e loading
I= s u p p o r t on b o t t o m
2= s u p p o r t on t o p
3- Fixed s u p p o r t
r i g h t s i d e support
l = l o a d on t o p
M A L Page A-83
Membrane-2000 Response-2000 Triax-2000 Shell-2000
2= load o n boztorn
3= fixed column base
[ PENETRATE -f num] yield penetration. 0.022 suggested
[ ENDMEMB ]
LKDINPUT
[more input-labels may be place here to as many as desired,]
Vecchio, F.J. and Collins, M.P.,"The Modified Compression Field Theory for
Reinforced Concrete Elements Subjected to Sheai', AC1 Journal, Proceedngs 83
No. 2, March-April 1986, pp. 219-23 1.
Collins, M.P. and Mitchell, D., "Prestressed Concrete Structures", Prentice-Hall 1991
760 pp.
KoIleger, J., Mehlhorn, G., "'Material Model for Cracked Reinforced Concrete,"
IABSE Colloquim on Computeational Mechanics of Concrete Structures: Advances
and plllications, Delft, 1987. Report No. 54, pp 63-74.
Mikame, A., Uchida, K., Noguchi, H., "A Study of Compressive Deterioration of
Cracked Concrete," Proceedings, Internationsl Workshop on FEA of RC, Columbia
University, New York, 1991.
15 Hsu, T.T.C, "Unified Theory of Reinforced Concrete", CRC Press, Inc, Boca Raton,
1993,336 pp.
16 Collins, M.P. and Mitchell, D., Prestresed Concrete Basics. Canadian Prestressed
Concrete Institute, 1987.
17 Izumo, J. Shin, H. Maekawa, K., Okarnura, H., "An analytical Model for RC Panels
Subjected to In-Plane Stresses," Concrete Shear in Earthquake, Elsevier Applied
Science, London and New York, 1992, pp.. 206-21 5.
18 Tamai, S. Shima, H., Izumo, J. and Okamura, H., "Average Stress-Strain Relationship
in Post Yield Range of Steel Bar in Concrete." Concrete Library of JSCE, No. 11,
June 1988, p. 117-129. (Translation fiom Proceedings of JSCE, No. 378N-6, Feb
1987).
19 AASHTO LRFD Bridge Design Specifications and Commentary, "First Ed.,
American Association of State Highway Transportation Officiais, Washington, 1194,
1901 pp. aashto code
20 Zhang, L-X, and Hsu, T.T.C., "Behavior and Analysis of 100 MPa Concrete
Membrane Elernents," Journal of Structural Engineering, ASCE, Vol 124 No. 1,
1998, pp 24-34.rastm-98
X.B.and Hsu, T.T.C. "Fixed Angle Softened Truss Model for Reinforced
2 1 Pang,
Concrete", ACIStmcturalJournaf,V 93, No 2., Mar. Apr. 1996, pp. 197-207
22 Hsu, T.T.C, and Zhang, L-X., 'Wonlinea. Analysis of Membrane Elements by Fixed-
Angle Softened-Tniss Model." AC1 Structural Journal, Vol. 94, No. 5, Sept-Oct 1997.
pp 483-492.
EAST
WEST
EAST
HS 1: Load
WEST
HS 1: Load Stage 5 and final state: North Face
HS2: Load Stage 1: North Face
TOP
WEST
HS2: Load
WEST
4
HS2: Load Stage 2: North Face
TOP
WEST
HS2: Load Stage 2: South Face
TOP
HS2: Load Stage 3: North Face
TOP
EAST WEST
O
EAST
HS2: Load Stage 4: North Face
TOe
WEST
HS2: Load Stage 4: South Face
T
WEST
HS2: Load Stage 6 and 7:North Face
HS2: Load Staee 6 and 7: South Face
HS2: Final State: North Face
HS2: Final State: South Face
Appendix C Experimental Verification Tables
for Response-2000
440 195.a
nm 398.9
me =.a
270 a).1
27Q W.8
270 75.7
m 71.2
270 Sd.3
270 m.1
270 56.8
270 56.6
270 20.0
270 24.5
270 25.4
270 25.1
270 57.0
270 3s.o
270 523
270 40.5
270 a 1
270 34.5
2-m 51.2
270 31 2
270 53.4
270 40.0
270 al1
270 47.2
270 54.6
270 43.2
270 512
270 37.8
278 1272
278 118.3
ne 107.9
m 139.3
278 m.1
278 65.6
278 48.5
278 46.7
278 85.6
178 74.3
378 118.9
278 m.0
278 81 .8
278 m.2
m 158.2
278 233.7
278 m.1
278 240.3
278 m.4
270 178.0
278 180.6
QSO a . 4
Q30 328.4
ms 101.3
&s 87.3
405 85.3
233 225
233 24.0
233 27.5
233 225
233 f1.0
233 27.5
1 s 11.6
1 s 121
1 s 10.6
139 11.4
on) 200.0
on1 m.7
on) 3120
910 315.8
445 es.6
445 m.5
445 m.5
on) m.2
on1 m.8
0a1 m.l
445 02.0
445 100.8
910 516.3
- Cwl
CW2
cw3
Cw4
CWS
CW7
cwo
cwo
CW8
ai
Ca3
Ct2
CU
as
a7
Cs
cm
CM
C w lO
cwl 1
CWl2
CW13
CW14
Cw15
m l 6
CW17
aiO
a 1i
a 12
a13
air
ais
Ci16
Cil 7
61
62
Dl
02
aco
Sc1
Sc2
SC3
SU
EB1
WB1
EB2
WB2
B l 00
BlOOR
Blao
BlOOH
Wlm
B i a.
BI WLd
BI OOB
ERLl00
BH0100
BH0100R
BHDSO
BHD50R
BHMS
en100
en50
w25
mol00
m050
BNMS
EN100
BN50
eu25
BMlOO loo0 3m
BMl000 lam 3m
WMIOOO lam low
WM10#: loo0 low
TrC- m
-7 3140 1-
shay.8 200 loo0
Shig5 1100 5m
ShDy. 3 ma rra
shpy.2 220 1 s
M a Po0 a0
A4b 2100 a0
A 38 1050 350
M b 1050 350
A-2 8 510 178
A-2 b m 178
A-: 8 330 105
A-1 b 330 105
=
3m
m
1a)
3m
=
la)
3a)
3m
1a)
la)
#O
rn
fa)
fa)
380
5a1
3m
=
ja)
flO
sa,
3m
3m
1a)
3a)
3m
371
371
371
37l
ni
371
371
371
371
ni
371
371
371
371
371
371
371
371
371
371
371
371
371
371
371
371
371
371
311
311
311
311
311
311
311
311
$11
311
311
311
Otl
953
Otl
933
033
a33
953
Rdrn 549
6 Russel 549
R d u a n c a 70 569
iaa #7
tao a.9
2Q.6 U.1
2l.3 39.2
2B.0 54.3
29.0 50.2
ZD.0 m.0
a2 9.5
21.0 54.3
21.3 S.1
79.6 S.6
a- 1 61.0
37.0 623
26.8 e.4
220 56.5
10.6 45.4
19.9 57.4
21.5 57.9
2l.o 52s
19.8 53.0
a).1 54.3
2l.4 53.4
23.4 w.2
18.8 63.6
25.0 423
t28 421
a-2 S.@
50.1 n.3
321 320
30.5 29.4
221 36.3
23.2 25.8
23.9 2x1
23.9 30.3
20.1 320
20.6 320
20.8 #.O
16.3 24.0
20.6 n.0
t27 R 5
46.4 82.7
10.0 31.7
21.9 34.3
19.8 3.6
20.3 35.6
10.8 3S.6
20.3 35.6
17.8 33.8
321 1
a).
19.3 33.8
2Q.o 34.3
37.4 61.4
16.7 34.7
225 56.2
226 9.9
30.8 55.6
23.2 47.2
2l.o 9.5
21.9 #. 1
2l.8 50.3
18.2 51.1
18.0 521
24.2 5d.5
24.1 9.7
44.3 46.7
20.0 23Q
21.9 23.6
229 81.O
27.8 673
H.5 1-9
#.S 1228.2
18.0 M.7
49.0 7m.Q
49.1 nY.5
a. 1 925.8
52.4 e2l.2
S24 1025.5
48.5 1143.7
50.3 1143.7
50.3 1m.4
524 417.9
524 434.3
51.4 m.0
51.4 @la7
I -?RN
N8n (mm)
1118
-2RS 1118
BOOOjRN 1118
m S 1118
m N 1118
COQlRs 1118
Cl40-RN 1118
Cl-O-RS 1118
C1-00-3W2N 1118
Cl-OKWRS 1118