Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

LAB REPORT 2

PREDICTIVE MODELLING

Submitted to:
Dr. Gordon Wight

Fei Xie
May 24, 2021
Table of contents
1. Introduction…………………………………………………………………………….. 1
2. Predictive Model Setup…………………………………………………….................... 1
3. Predictive Model Results …………………………………………………..................... 1
4. Experimental Model Results …………………………………........................................ 3
5. Discussion ……................................................................................................................ 5
6. Summary …...................................................................................................................... 5
Annex
Figure No. Figure Description Page No.
1. Beam 1 predictive model load-deflection curve. 2
2. Beam 4 predictive model load-deflection curve. 2
3. Beam 8 predictive model load-deflection curve. 2
4. Beam 9 predictive model load-deflection curve. 3
5. Beam 1 experimental model load-deflection curve. 3
6 Beam 4 experimental model load-deflection curve. 4
7. Beam 8 experimental model load-deflection curve. 4
8. Beam 9 experimental model load-deflection curve. 4
1. Introduction
A predictive model can be used to show the flexural behaviour of test beams. The model is built
upon strain compatibility approach, the equilibrium of forces and constitutive material properties.
A moment-curvature relationship can be obtained through VBA Macro to model the behaviour of
the beams and then turned into a load-deflection curve to compare to the results from the
experimental tests.
A total of nine beams with different steel reinforcement and FRP strengthening were tested in lab
and the experimental data were recorded. Furthermore, the predictive models of beams 1, 4, 8, and
9 were built via VBA to model their load-deflection behaviours and compare with the experimental
results. Beam 1 has two size 10M steel rebars and beam 4 has three size 10M steel rebars for
flexural reinforcement, both with no CFRP strengthening. Beam 8 has the same steel reinforcement
as beam 1, and an extra four layers of CFRP at the bottom. Beam 9 also has two size 10M steel
reinforcements and surrounded four layers of CFRP strengthening.

2. Predictive Model Setup


The tested beams are cuboids, with 1800 mm length,150 mm width and 200 mm depth. The steel
yield strength is assumed to be 400 MPa. The thickness of each CFRP layer is 0.167 mm. To build
the predictive model, the depth is evenly divided into twenty pieces of 10mm thick slabs. The
strains of steel, and CFRP are assumed to be the same as the concrete within the same slab. For
each slab, the strain of concrete is in linear relationship with the top and bottom concrete strains.
The top concrete strain ranges from -0.0035 (compression) to 0.001 (tension). Although concrete
compressive strain can reach up to 0.004 in this experiment, 0.0035 is normally considered as the
maximum compressive strain that can be achieved in concrete. In addition, the sum of forces for
concrete, steel and CFRP in all slabs should be equal to zero because of the equilibrium of forces.
After that, the bottom strains regarding different top concrete strains are calculated via VBA Macro
where the sum of forces is set to be zero. Moment and curvature are also calculated and recorded
along with the bottom strain.
To get the predictive load-deflection curve, moment arms from support (0 m) to midspan (0.9 m)
are applied with 0.1 m increment. Different curvature values can be achieved at various moment
arms for increasing loads. Furthermore, the deflection at midspan can be calculated using the
equation approximated by Collins and Mitchell (1997). The maximum moment obtained through
VBA Macro is also the moment limitation here. Eventually, the load-deflection curves for
predictive models are plotted and compared to the experimental model curves.

3. Predictive Model Results


The predictive load-deflection curves of all four beams considered are shown in this section. These
beams are modeled until the smallest of their corresponding maximum moments and experimental
load is reached.

1
Figure 1. Beam 1 predictive model load-deflection curve.

Figure 2. Beam 4 predictive model load-deflection curve.

Figure 3. Beam 8 predictive model load-deflection curve.

2
Figure 4. Beam 9 predictive model load-deflection curve.

4. Experimental Results
The experimental load-deflection curves of all four beams considered are shown in this section.
These experimental curves are plotted that the same deflection range is achieved for both
experimental and predictive models.

Figure 5. Beam 1 experimental model load-deflection curve.

3
Figure 6. Beam 4 experimental model load-deflection curve.

Figure 7. Beam 8 experimental model load-deflection curve.

Figure 8. Beam 9 experimental model load-deflection curve.

4
5. Discussion
For beam 1, the maximum predictive moment achieved via VBA Macro is 13.25 kNm when the
top concrete strain is 0.0035 in compression. The maximum moment occurs at midspan,
corresponding to a maximum deflection of 13.15 mm. The load-deflection curves for both
experimental and predictive models are quite similar except that the experimental curve
approaches a maximum load of 55 kN at maximum deflection whereas the predictive model
reaches 50 kN. This is mainly because the yield strength of steel used in the predictive model is
400 MPa, but the actual yield strength of the steel reinforcement is higher than 400 MPa. Overall,
the difference in yield strength does not have a great impact on the predictive model.
Beam 4 reaches the maximum predictive moment of 48.5 kNm when the top concrete strain is at
0.0035 in compression. This maximum moment corresponds to a load of 176 kN at midspan for
the predictive model. However, the actual maximum load achieved in the experiment was 112 kN,
which leads to a maximum deflection of 3.27 mm in the predictive model. In general, both the
predictive and experimental curves exhibit a linear relationship between load and deflection. It is
worth mentioning that the slope of the predictive load-deflection curve is steeper than that of the
experimental curve, which means that a higher load is required for the predictive model than the
actual load to reach the same deflection. This is counterintuitive because the yield strength of steel
used in the predictive model is lower than the actual yield strength, so the slope should be smaller.
The general load-deflection behaviour of beam 4 is still perfectly represented by the predictive
model.
For beam 8, the maximum predictive moment achieved via VBA Macro is 24.8 kNm when the top
concrete strain is 0.0014 in compression. This maximum moment occurs at midspan,
corresponding to a maximum deflection of 4.65 mm. In general, the predictive load-deflection
curve perfectly matches the experimental load-deflection behaviour, these two curves look quite
similar. However, the same problem occurs for both beam 4 and beam 8 predictive models. A
higher load is required for the predictive model than the actual load to reach the same deflection.
Therefore, a discussion is made with Denis and the conclusion is that there is lack of sufficient
shear reinforcement which has a great impact on the load-deflection behaviour. The shear
reinforcement has a significant impact on the load-deflection behaviour and cannot be neglected.
Beam 9 reaches the maximum predictive moment of 18.4 kNm when the top concrete strain is at
0.0035 in compression. This maximum moment corresponds to a maximum deflection of 7.82 mm
at the load of 67 kN for the predictive model. In general, the predictive load-deflection curve
perfectly reflects the load-deflection behaviour of the actual beam 9. A little larger load is required
for the predictive model than the actual beam to achieve the same deflection, but it does not affect
the general behaviour and is considered reasonably acceptable.

6. Summary
In summary, the predictive models can clearly reflect the actual load-deflection behaviours of the
actual beams. Although customized parameter values, shear reinforcements, the selected top
concrete strain range, and the size of load increment may lead to the difference in the load-

5
deflection behaviour between predictive model results and experimental model results, the overall
relationship is still accurately exhibited. Therefore, the Macro VBA is proved to be an effective
way of building a predictive model.

You might also like