Sie sind auf Seite 1von 39

CIVL 821: Finite Element Method

Queens University, Canada

Load-Deflection Response of a
Prestressed Concrete T-Beam

Prepared for:
Dr. Moore

Prepared by:
Ray Saiedi

December 2007
CONTENTS

1. PROBLEM STATEMENT...................................................................................................................... 1
1.1 Introduction ................................................................................................................................... 1
1.2 Objectives....................................................................................................................................... 1
1.3 Experimental Program................................................................................................................... 2
2. MODELING ISSUES............................................................................................................................... 4
2.1 ASSUMPTIONS................................................................................................................................. 4
2.2 GEOMETRY ...................................................................................................................................... 5
2.2.1 Coordinate system....................................................................................................................... 5
2.2.2 Symmetry..................................................................................................................................... 5
2.2.3 Simplification of Model............................................................................................................... 6
2.2.4 Cross Section .............................................................................................................................. 7
2.3 BOUNDARY CONDITIONS ............................................................................................................. 8
2.3.1 Transverse Half........................................................................................................................... 8
2.3.2 Longitudinal Half........................................................................................................................ 9
2.3.3 Roller Support........................................................................................................................... 10
2.4 LOADS ............................................................................................................................................. 12
2.4.1 Loading Area ............................................................................................................................ 12
2.4.2 Load Cases................................................................................................................................ 13
2.5 MATERIALS.................................................................................................................................... 14
2.5.1 Model Excludes Prestressing Steel ........................................................................................... 14
2.5.2 Modeled Concrete Can Resist Tension ..................................................................................... 15
2.5.3 Material Properties.................................................................................................................... 16
2.6 MESH ............................................................................................................................................... 17
2.6.1 Elements.................................................................................................................................... 17
2.6.2 Mesh Layout.............................................................................................................................. 17
2.6.3 Paths ......................................................................................................................................... 19
3. RESULTS & DISCUSSION ................................................................................................................. 20
Shear & Moment Diagrams ................................................................................................................ 21
3.1 LOAD VS. DEFLECTION ............................................................................................................... 22
3.1.1 Results from Model ................................................................................................................... 22
3.1.2 Results from Tests ..................................................................................................................... 25
3.1.3 Model vs. Tests.......................................................................................................................... 26
3.1.4 Model vs. Theory....................................................................................................................... 27
3.2 LOAD AT YIELDING...................................................................................................................... 28
3.2.1 Results from Model ................................................................................................................... 28
3.2.2 Result from Tests....................................................................................................................... 32
3.2.3 Model vs. Tests.......................................................................................................................... 32
4. APPENDICES ....................................................................................................................................... 33
A: AutoCAD drawing of beam cross section....................................................................................... 34
B: Excerpt from CISC Steel Handbook for manual deflection calculations ....................................... 35
C: Manual calculations of stress distribution due to prestress........................................................... 36
D: PCI Journal Article: Long-Term Behavior of CFRP Prestressed Concrete Beams ...................... 37

ii
1. Problem Statement
This chapter introduces the experimental problem for which a finite element analysis is to
be performed, outlines the objectives of this analysis, and summarizes the geometry,
materials and loading setup to be modeled.

1.1 Introduction

The journal article enclosed in Appendix D presents an experimental program designed to


examine the long-term behavior of carbon fiber reinforced polymer (CFRP) pretensioned
concrete beams under sustained load conditions. As part of the study, however, the
short-term behavior of a steel pretensioned concrete beam was examined during initial
static loading to the sustained load level. Fig. 7 on page 103 of the article presents the
load-deflection response of the beam for this study obtained via lab tests. Figure 1.1
shows a simplified version of that figure (only the curve corresponding to the steel
pretensioned concrete beam).

60

50

40
Load (kN)

30

20

10

0
0 1 2 3 4 5 6 7 8 9 10
Deflection (mm)

Figure 1.1: A Simplified reproduction of beams load-deflection response from lab tests
Extracted from Fig. 7 of Appendix D

1.2 Objectives

The objective of this project is to perform an independent check of the results in this
graph using finite element analysis. This will validate the experimental results against
the finite element model. Specifically, the project aims to predict:

a) the load-deflection response in the elastic region of the graph


b) the load at yielding (where the beam ceases to behave elastically)

1
1.3 Experimental Program

Figure 1.2 shows the cross section of the reinforced prestressed concrete beam. MK1A
designates the prestressing strands. MK2, MK3 and MK4 are steel rebar, steel stirrup
and reinforcing steel plate, respectively.

Figure 1.2: Beam cross section and reinforcement details

The beam was prestressed with two prestressing steel strands to 202 kN. The concrete
was designed to have a 28-day compressive strength of 40 MPa.

Figures 1.2 and 1.3 show the loading arrangement for the 4 beams tested in the lab. The
beams were simply supported on concrete blocks and subjected to third-point loading in a
load frame. The load was applied by stressing two steel strands vertically between a
hollow steel section and two steel channels anchored to the laboratory strong floor. The
hollow steel section reacted on a spreader loading beam that in turn applied concentrated
loads at third points of the prestressed beam.

Figure 1.3: Photo of loading frame

2
Figure 1.4: Schematic showing loading frame

3
2. Modeling Issues
This chapter describes all the considerations made in creating the finite element model.
These considerations have been categorized into the following categories: assumptions,
geometry, boundary conditions, loads, materials, and the mesh.

2.1 ASSUMPTIONS

We shall assume the following:

1. The roller support reacts on one transverse line under the beam rather than on an
small intermediary bearing plate. See Section 2.3.3 for details.

2. In Figure 1.4 loading is completely symmetric on the beam both longitudinally


and transversely. Therefore:

a) The spreader beam distributes load equally among the two lines of loading.
b) Each transverse half of the beam on the two lines of loading receives equal
loading.

3. Loading is applied uniformly over a rectangular area 100 mm wide and 130 mm
long, centered on the beam transversely. See Section 2.4.1 for details.

4. Concrete is isotropic and uniform throughout the beam.

5. The concrete in the experimental beam cannot resist any tension. It will crack
under the smallest tensile stress.

6. The concrete in the model beam can resist tension just as well as it can resist
compression. This assumption is justified in Section 2.5.2.

7. There is no loss of prestress in the beam, therefore the prestressing force at every
point in the beam is equal to the jacking force.

8. Vertical deflection in a beam is independent of the amount of prestressing force


present in it. This assumption is not strictly true. For example, Table 5 in the
article in Appendix D shows that these two parameters are in fact dependent.
However, their dependency is very small and will be neglected in this study.

4
2.2 GEOMETRY

2.2.1 Coordinate system

Figure 2.1 shows the Original Beam modeled in 3D. The x, y and z axes point in the
transverse, vertical and longitudinal directions of the beam, respectively. Although not
shown in this figure, the origin of our 3D model was chosen to be at the bottom edge of
the web, centered transversely and longitudinally in the beam.

Figure 2.1: Orientation of x, y, z axes

2.2.2 Symmetry

As seen in Figure 2.2, the beam cross section is symmetrical about the vertical centerline.
Loading is also transversely symmetrical. Thus, the plane that cuts the beam
longitudinally and includes the axis of symmetry of the cross-section (the y-z plane) is
a plane of symmetry for the 3D beam.

Figure 2.2: Cross section of Original Beam

5
From Figure 1.4 it is evident that the beams geometry is also symmetrical longitudinally.
So is the loading applied to it, i.e. the loading beam applies equal loads at points 1,333 mm
apart from each other and 1,333/2 = 666.5 mm away from the longitudinal mid-point of
the beam. Thus, the plane that crosses the beam transversely at the mid-point of its
length (the x-y plane) is also a plane of symmetry.

2.2.3 Simplification of Model

Given the double-symmetry of the beam and the applied loads, it is possible to halve the
beam transversely and longitudinally, and model only a quarter of the beam instead of the
entire beam. See Figure 2.3.

Figure 2.3: The modeled beam is a quarter of the Original Beam

The advantages of this approach are:

a) On the users part: simplification of the problem and reduced modeling effort
b) On the computers part: a finer mesh (same number of nodes on a smaller
volume) and thus, more accurate results

As a consequence, we will have only one longitudinal line of loading rather than two
lines in the Original Beam. Also, as the original top flange was halved transversely, only
one half of the load exerted on this line will be applied to our beam. Therefore, the load
applied to the beam will be a quarter of the original load: F = F / 4

Throughout the remainder of this report:

Original Beam refers to the 4.4m-long, doubly symmetric beam on which lab
tests were performed.
Beam and Quarter-Beam refer to the new 2.2m-long beam that is a quarter of
the Original Beam and on which finite element analysis is performed.

6
2.2.4 Cross Section

The beam was defined as a 3-D part in ABAQUS. First a sketch of the beam cross-
section was drawn in 2-D as in Figure 2.4. All dimensions were according to Figure 1.2
which was extracted from the article in Appendix D.

Figure 2.4: Two-dimensional sketch of beam cross section

Table 2.1 contains the coordinates of the points in Figure 2.4.

Table 2.1: Coordinates of points in Figure 2.4


Point X (m) Y (m)
A 0.000 0.000
B 0.065 0.000
C 0.0775 0.250
D 0.250 0.250
E 0.250 0.300
F 0.000 0.300

This sketch was made 3-dimensional by giving it a depth equal to half the length of the
Original Beam: L = 4.4 m /2 = 2.2 m

Appendix A shows an AutoCAD drawing of the cross section. It also includes


automatically calculated section properties such as area (A), location of neutral axis ( y ),
and moment of inertia (Ix) which we will later use in our manual calculations.

7
2.3 BOUNDARY CONDITIONS

2.3.1 Transverse Half

Recall that the y-z plane is a plane of symmetry for the Original Beam and for the loading
arrangement. Naturally, when the beam is loaded, there will be no transverse
displacements across this plane. This plane is said to act as a smooth rigid boundary:

It is smooth in that it allows displacement in the vertical (y) and longitudinal (z)
directions.
It is rigid in that it does not allow displacement in the transverse (x) direction.

Therefore it can be modeled as a plane with a very simple boundary condition: Restrict
U1. Figure 2.5 shows this boundary condition in red.

Figure 2.5: Boundary condition on transverse half of beam

8
2.3.2 Longitudinal Half

Recall that the Original Beam and the loading arrangement are also symmetrical about
the x-y plane. Then this plane also acts as a smooth rigid boundary. It can be modeled as
a plane with the boundary condition: Restrict U3. Figure 2.6 shows this boundary
condition in red.

Figure 2.6: Boundary condition on longitudinal half of beam

9
2.3.3 Roller Support

The roller support was introduced by partitioning the bottom face of the beam at the point
of support and imposing the boundary condition of restraint against vertical displacement
(U2) on the edge between the partitions. See Figures 2.7 and 2.8.

Figure 2.7: Partition line at location of roller support (view of beam from underneath)

Figure 2.8: Location of roller support marked in red

The precise coordinates of the roller support are:

x: From 0.000 to 0.065 m, y = 0.000, z = 2.000 m

10
In the laboratory tests, a bearing plate was placed between the roller and the beam in
order to spread the vertical reaction on a wider contact area and thus prevent
concentration of local stresses in the concrete.

In order to model this in finite element software one would have to impose the boundary
conditions of restraint against vertical displacement on the area of this rectangular plate
rather than on an edge located at its longitudinal centerline. However, doing so would
restrict the beam from rotating longitudinally. Consequently, the support would not be a
roller support anymore. For this reason, it was decided to continue using the edge
support instead.

As a consequence of this inaccurate representation, local stresses at the support will be


exaggerated. This is observed on the left hand side of Figure 2.9. These stresses should
be ignored.

Figure 2.9: Stress contours show exaggerated local stresses at location of roller support

11
2.4 LOADS

2.4.1 Loading Area

In the laboratory tests, a small bearing plate was placed between the loading beam and
the concrete beam. This plate spreads the external force to a wider contact area on the
concrete beam, thus preventing concentration of stress in the concrete beam at the point
of contact. This effect of the bearing plate was simulated in the model by defining the
external force not as a concentrated load, but as a pressure load on a contact area the size
of the bearing plate. Figure 2.10 shows how this contact area was created by partitioning
the top flange in plan view.

Figure 2.10: Plan view of top flange and defined loading area

Figure 2.11 shows this loading area in 3D, marking the pressure load applied on it in red.

Figure 2.11: Pressure on loading area

12
The coordinates of the loading surface are:

X: From 0.000 to 0.065 m, Y = 0.300 m, Z: From 0.6165 to 0.7165 m

Thus, for every new value of load to be applied to the Original Beam (F), the pressure to
be applied to the contact area on the quarter-beam in our model (P) is calculated as
follows:

A = 0.1( m ) 0.065 ( m ) = 0.0065 ( m )


2

F ( kN ) F ( kN ) / 4 F ( kN )
P ( kPa ) = ( m 2 ) = 2
= 2
A 0.0065 ( m ) 0.026 ( m )

2.4.2 Load Cases

The 1st objective of this study is to predict the load-deflection response of the beam in the
elastic region of Figure 1.1. Visual inspection determines that this region ends at a load
of 32.9 kN. Thus, we shall define the following load cases:

1. F = 10kN
2. F = 20kN
3. F = 30kN

The advantage of choosing 3 points is that it allows us to judge if the load-deflection


response is linear or not.

The 2nd objective of this study is to find the load at yielding. As we will see later, this
will be determined by interpolation from the results of F = 10, 20, 30 kN. So no
additional load cases are necessary.

13
2.5 MATERIALS

2.5.1 Model Excludes Prestressing Steel

It was decided that prestressing steel not be included in our computer model, because:

Doing so will complicate the model beyond the scope of this project.
Doing so will affect the results of our analyses hardly at all. This is because
prestressing steel contributes very little to the stiffness and resulting deflections in
our beam. This is demonstrated theoretically below:

The stiffness of a prestressed beam is found as follows:

1. Prestressed beams are composed of two members: concrete and prestressing steel.
2. In composite sections, stiffnesses of individual members can be summed to arrive
at the stiffness of the whole.
3. The stiffness term, EI, is composed of Modulus of Elasticity, E and Moment of
Inertia, I.

Therefore, Eb . Ib = Ec . Ic + Es . Is where subscripts c, s and b represent concrete,


prestressing steel, and the prestressed concrete beam, respectively.

Although the modulus of elasticity of steel strands is about 7 times greater than concrete,
their cross-sectional area and therefore their moment of inertia is relatively very small.
The net result is a stiffness that is small enough to be neglected:

Concrete: Prestressing Steel:


Ec = 28,460 MPa Es = 200,000 MPa
I c = 252.21 10 6 mm 4 r 4 (13 / 2) 4
Is = = = 1.4 10 3 mm 4
Ec I c = 7,177,897 N.m2 4 4
E s I s = 280 N.m2

Prestressed Beam:
Eb . Ib = Ec . Ic + Es . Is = 7,177,897 + 280 = 7,178,177

Es I s 280
= = 0.003%
E b I b 7,177,897

Consequently prestressing steel contributes only 0.003% to the stiffness of the beam and
excluding it will not cause noticeable changes in the results.

14
2.5.2 Modeled Concrete Can Resist Tension

In the model, the beam was given a modulus of elasticity equal to that of concrete.
However, it was defined as being capable of resisting both compression and tension,
whereas real concrete can only properly resist compression; its tensile resistance is very
low. This section will explain why this difference does not cause problems for the
purposes of this study.

If a prestressing force did not exist, the concrete in the test beam would crack under
tension and the beam would fail completely under small loads. Our model, however,
would assume that concrete can resist tension and the beam would remain intact. The
model would thus grossly misjudge the actual behavior of the beam.

The force and location of the prestressing steel are designed such that when there is no
load the entire web is put under compression. The writers calculations show that even
under the greatest load for which the relationship between load and deflection is being
studied (30kN), the entire web remains in compression. That is, the tension produced in
the bottom fibers of the web is not enough to cancel out the compression. Since there is
no tension in the beam, it is safe to assume concrete can resist tension.

In order to find the stress distribution in a prestressed beam under load, we superimpose:
a) the stress distribution due to prestress, and
b) the stress distribution due to the applied load (consisting of both positive and
negative stresses).

In this superimposition, the negative stresses due to applied load in fact do not ultimately
result in tensile stresses, they merely reduce the pre-existing compressive stresses created
by the prestress. The deflections resulting purely from applied load in this situation are
therefore the same as those obtained by assuming the beam has no prestress but the
concrete can resist both compressive and tensile stresses.

15
2.5.3 Material Properties

The materials used in the Original Beam include concrete, reinforcing steel and
prestressing steel. However, the only material used in the model is concrete.

The article in Appendix D does not directly give the modulus of elasticity (E) of the
concrete used to pour the beams. However, it states its compressive strength (fc) is 40
MPa. Thus, Ec can be calculated as:

E c ( MPa) = 4500 f c ( MPa)


E c = 4500 40 = 28,460 MPa = 28,460 10 3 kPa

Table 2.2 summarizes the material properties used in the model.

Table 2.2: Material properties used in model


Material Modulus of Elasticity, E (kPa) Poissons Ratio,
Concrete 28,460 E+3 0.15

16
2.6 MESH

2.6.1 Elements

Generally compact and regularly shaped elements give the most accurate results.
Therefore, tetrahedral elements were chosen as they are the most compact and regularly
shaped 3D elements available. The family of elements was chosen as 3-dimensional
stress analysis.

2.6.2 Mesh Layout

The academic license of ABAQUS available to the writer restricts the number of nodes to
less than 20,000. Therefore, in order to make the most of the nodes available it was
necessary to have a more refined mesh in regions of high strain or stress gradients. In
areas of slow variation relatively large elements will usually yield sufficient accuracy
while in areas of rapid variation much smaller elements are required to capture the
solution.

To find areas of high stress gradients, a preliminary mesh was defined with elements
having the same size and shape throughout the continuum. An analysis of the stress
distribution showed the stress variation to be highest through the depth of the beam
between the loading point and the mid-point of the Original Beam. Stresses changed
slowly at the end of the beam near the support. See Figure 2.12.

Figure 2.12: Stress contours from mesh with equally-sized elements

17
Hence, in the final mesh, for those edges between the point of loading and the mid-point
of the Original Beam, seed size was chosen as 10 mm. For the portion of the beam
between the loading point and the end near the support, a biased mesh was defined with a
bias ratio of 5:1, such that seeds would start at 10 mm at the point of loading and would
end at approximately 50 mm at the end near the support. Figure 2.13 shows the seeds on
the edges. Figure 2.14 illustrates the mesh produced based on those seeds.

Figure 2.13: Seeds defined on beam edges

Figure 2.14: Mesh created from defined seeds

18
2.6.3 Paths

A Path in ABAQUS is a series of nodes for which different variables can be readily
plotted. Paths allow the user to quickly identify variable changes along given boundaries.
This approach is much simpler than extracting numeric data to a text file and then
manually plotting them in a spreadsheet.

For the purposes of this assignment, two paths were defined:

1. Longitudinal Path: This consisted of nodes at the very top of the top flange on
the edge directly above the longitudinal centerline of the original web. The path
starts at the end closest to the roller support and progresses towards the middle of
the Original Beam. Figure 2.19 marks this path in red.

Figure 2.19: Longitudinal Path

2. Vertical Path: This consists of nodes along the vertical line located at the
transverse and longitudinal midpoint of the Original Beam. The path starts at the
top of the beam and ends at the bottom. Figure 2.20 displays this path in red.

Figure 2.20: Vertical Path

19
3. Results & Discussion
This chapter presents the results of the finite element analysis on the computer model and
compares them with:

a) results obtained from lab tests on the beam, and


b) analytical results obtained from engineering theory.

It then attempts to justify the discrepancies. This process allows us to verify whether the
finite element analysis was used correctly.

As per the projects objectives in Section 1.2, the results focus on two main themes:

1. Load vs. Deflection


2. Load at Yielding

20
Shear & Moment Diagrams

As a first step, let us draw the shear force and bending moment diagrams of the Original
Beam for the applied load of 30 kN. See Figures 3.1 and 3.2. These diagrams will be
used in later discussions.

20

15

10
Shear, V (kN)

5
0.667 2
0 2.2
-2.2 -2 -0.667 0
-5

-10

-15

-20
Longitudinal distance, Z (m)

Figure 3.1: Shear force diagram of Original Beam for 30 kN loading

25
Bending Moment, M (kN.m)

20
-0.667 0.667
15

10

0
-2.2 -2 0 2 2.2
Longitudinal distance, Z (m)

Figure 3.2: Bending moment diagram of Original Beam for 30 kN loading

21
3.1 LOAD VS. DEFLECTION

3.1.1 Results from Model

Figure 3.3 shows the vertical displacements along the Longitudinal Path for 30 kN
loading. These vertical displacements are color-coded in Figure 3.4 for all points of the
beam. As expected, vertical displacement is positive before the roller support, is zero at
the roller support, and is negative after it. Vertical displacement increases in magnitude
almost linearly from the roller support to the point of loading (from left to right on the
graph) and then at a gradually decreasing rate until the mid-point of the Original Beam
(at the right-most end of the graph).

Figure 3.3: Vertical displacement along Longitudinal Path for 30 kN loading

Figure 3.4: Vertical displacement contours in beam

22
Figures 3.5 and 3.6 show similar graphs for 20 kN and 10 kN loading, respectively.

Figure 3.5: Displacement along Longitudinal Path for 20 kN loading

Figure 3.6: Displacement along Longitudinal Path for 10 kN loading

23
The vertical deflections at the mid-point of the Original Beam (the right-most end of the
graphs) were extracted for varying levels of applied load. These are summarized in Table
3.1 and then plotted in Figure 3.7.

Table 3.1: Load vs. Deflection from Model


Load (kN) Deflection (mm)
0 0
10 0.82
20 1.64
30 2.46

35

30 2.46, 30

25
Load (kN)

20 1.64, 20

15

10 0.82, 10

0 0, 0
0 0.5 1 1.5 2 2.5 3
Deflection (mm)

Figure 3.7: Load vs. Deflection from Model

24
3.1.2 Results from Tests

Fig. 7 of the article in Appendix D shows the load-deflection response obtained from
laboratory tests. Figure 3.8 reproduces only the curve corresponding to the beam
pretensioned using steel strands and includes data labels for measured points.

60

50 9.84, 50

40 6.25, 45
4.56, 40
Load (kN)

3.32, 32.9 3.61, 35


30 3.01, 30

20 1.95, 20

10 0.89, 10

0
0.00 2.00 4.00 6.00 8.00 10.00 12.00
Deflection (mm)

Figure 3.8: Load vs. Deflection from Tests

25
3.1.3 Model vs. Tests

Figure 3.9 plots the load deflection curves from the model and the lab tests together in
one graph. Note that the curve associated with lab tests was only drawn up to the point of
yielding. A comparison of the two curves reveals a close match between the two. Table
3.2 lists the percentage of error for each load case.

50

40
Load (kN)

30 2.46 3.01
From Model
From Tests
20 1.64 1.95

10 0.82 0.89

0 0
0.00
0 0.5 1 1.5 2 2.5 3 3.5 4
Deflection (mm)

Figure 3.9: Load vs. Deflection from model and tests

Table 3.2: Load vs. Deflection from model and tests


Load (kN) Deflection (mm)
From Model From Tests Error (%)
10 0.82 0.89 9
20 1.64 1.95 19
30 2.46 3.01 22

It is seen that the model consistently underestimates the deflections obtained from
lab tests with errors of up to 22%. This means that the actual beam is weaker and less
stiff than the modeled beam. These errors can be explained as follows:

According to the article in Appendix D, the beam was almost 2 years old when the tests
were performed. During this time the beam was resting in the Structures Lab of the Civil
Engineering Department at Queens University. The writer asked a senior technician in
the Lab about the history of the beams and was told that during those 2 years:

The beam had been moved around by crane every now and again to make space
for other specimens. The beam may have been damaged during these movements.
The beam has periodically been loaded with several concrete cubes with a total
weight of 24 kN. Such long-term loading of the beam may have resulted in creep
and permanent damage to the beam.

26
3.1.4 Model vs. Theory

As an additional check for the results of our finite element analysis, we shall also obtain
the deflection at the middle of the beam using theoretical methods.

Appendix B is an excerpt from the CISC Handbook of Steel Construction. Section 9 in


this excerpt provides a formula for calculating deflection at the centre of a simply
supported beam loaded by two equal concentrated loads placed symmetrically:

Pa
max (centre) = (3l 2 4a 2 )
24 EI

For the load case of 30 kN, we have:

P = F/2 = 30/2 = 15 kN = 15,000 N


a = 1,333 mm = 1.333 m
E = 28,460 MPa = 28,460 10 6 Pa
I = 2 252.21 10 6 = 504.42 10 6 mm4
l= 4m

Thus:
15,000 1.333
max (centre) = 6
(3 4 2 4 1.333 2 ) = 4.74 10 3 m = 2.37mm
24 28,460 10 504.42 10
6

The analytically obtained deflection of 2.37 mm tallies closely with the result of our
finite element analysis, 2.46 mm, for a load of 30 kN. The error is less than 4%.
Therefore, our finite element analysis was correct. Any inconsistencies with
laboratory results must have an explanation beyond the realm of theory.

27
3.2 LOAD AT YIELDING

3.2.1 Results from Model

3.2.1.1 Stress Distribution in Cross Section

Figure 3.10 shows the longitudinal stress along the Vertical Path for 30 kN loading. The
Vertical Path starts at the top of the flange and ends at the bottom of the web. Negative
stress indicates compression while positive stress signifies tension.

Figure 3.10: Stress along Vertical Path for 30 kN loading

It is seen that stress varies linearly through the depth of the beam, from high compression
in the top of the flange to high tension at the bottom of the web. Because the flange at
the top is wider than the web beneath it, the neutral axis lies above the vertical midpoint
of the beam as indicated in Appendix A (189 mm > 150 mm). Therefore, the tensile
stress at the bottom of the web is greater in magnitude than the compressive stress at the
top of the flange.

This same stress distribution is color-coded in the stress contours of Figure 3.11. The
pattern of stress varying through the depth of the cross-section is seen to be strongest
between the loading point and the mid-point of the Original Beam. The combination of
tension at the bottom and compression at the top helps resist the high bending moment
induced in this region, illustrated in the bending moment diagram in Figure 3.2.

28
Figure 3.11: Contours of longitudinal stress in beam

3.2.1.2 Stress vs. Applied Load

Figures 3.12 and 3.13 show similar graphs to Figure 3.10 for 20 kN and 10 kN loading,
respectively.

Figure 3.12: Stress along Vertical Path for 20 kN loading

29
Figure 3.13: Stress along Vertical Path for 10 kN loading

The longitudinal tensile stresses at the very bottom of the web (the right-most end of the
graphs) were extracted for varying levels of applied load. These values are summarized
in Table 3.3 and then plotted in Figure 3.14. It is clear from this figure that load and
stress have a linear relationship. This linearity will be used in section 3.2.1.4 to find the
load at yielding from the model.

Table 3.3: Load vs. Stress from model


Load (kN) Longitudinal Stress (MPa)
10 2.584
20 5.163
30 7.748

40

30 7.748
Load (kN)

20 5.163

10 2.584

0 0
0 1 2 3 4 5 6 7 8 9
Stress (MPa)

Figure 3.14: Load vs. Stress from model

30
3.2.1.3 Yielding Mechanism

The 2nd objective of this study was to predict using our model the load at yielding, i.e. the
load at which the beam ceases to behave elastically. To do this we must return to the
basic principles of prestressed concrete.

The beam will only stop behaving elastically once the concrete starts to crack and gaps
begin to form. As stated earlier in section 2.5.2, concrete is very weak in tension. In
most analytical studies it is conservatively assumed that concrete will crack as soon as it
is put under tension. Thus, the beam will lose elasticity once the concrete is in tension.

Fortunately, the prestress applied to the beam keeps the entire cross-section in
compression before any load is applied. Imposed downward loads on the beam then tend
to reduce the compressive stress in the bottom extremities of the web. As long as there is
still some compression left in that area after load is applied, the concrete will remain
intact and the beam will remain elastic.

3.2.1.4 Load at Yielding

The load at yielding is therefore that load which produces a tensile stress at the bottom
fibers of the web equal in magnitude to the compressive stress that pre-exists there due to
prestress. This tensile stress will cancel out the compressive stress. Any further tension
will cause the concrete to crack and the beam to lose its elasticity.

Appendix C shows the procedures in calculating this critical stress. It was finally found
to be 11.974 MPa.

As concluded in section 3.2.1.2, applied load and longitudinal stress are linearly
proportional. Thus, we may use linear interpolation to find the load that would cause a
stress of exactly 11.974 MPa:

Load (kN) Longitudinal Stress (MPa)


0 0
30 7.748
X 11.974

11.974 30
X = = 46.36 kN
7.748

Therefore, the load at yielding obtained from the model is 46.36 kN.

31
3.2.2 Result from Tests

Figure 3.7 shows the load-deflection relationship obtained from lab tests and is re-
presented here in Figure 3.15. The load at yielding is the load at which the curve ceases
to be linear, that is, where the slope of the line changes. This is estimated by visual
inspection to happen at 32.9 kN.

60

50 9.84, 50

40 6.25, 45
4.56, 40
Load (kN)

3.32, 32.9 3.61, 35


30
3.01, 30

20 1.95, 20

10 0.89, 10

0
0.00 2.00 4.00 6.00 8.00 10.00 12.00
Deflection (mm)

Figure 3.15: Load vs. Deflection from Tests

3.2.3 Model vs. Tests

The model predicted the load at yielding to be 46.36 kN while the tests measured it as
32.9 kN. Once again, our model has overestimated the strength of the beam. The
46.36
error in the models prediction is ( 1) 100% = 41%.
32.9

This is a significant error. However, it can be justified in the following manner:

The hand calculations in Appendix C assume that the prestressing force in each strand at
the centre of the Original Beam is equal to the jacking force (see Assumption 7 in section
2.1). In reality, however, much of the jacking force is lost during and after prestressing.
This is termed loss of prestress and occurs through mechanisms such as anchorage
draw-in, elastic shortening, creep of concrete, shrinkage of concrete, and steel relaxation
in the prestressing strands. Losses of 20 - 40% are typical. Taking into account
prestress losses in our model would reduce the estimated load at yielding, bringing it
within acceptable range of the test results.

32
4. Appendices

A: AutoCAD drawing of beam cross section

B: Excerpt from CISC Steel Handbook for manual deflection calculations

C: Manual calculations of stress distribution due to prestress

D: PCI Journal Article: Long-Term Behavior of CFRP Prestressed Concrete Beams

33
A: AutoCAD drawing of beam cross section

34
B: Excerpt from CISC Steel Handbook for manual deflection calculations

35
C: Manual calculations of stress distribution due to prestress

36
D: PCI Journal Article: Long-Term Behavior of CFRP Prestressed Concrete Beams

37

Das könnte Ihnen auch gefallen