Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Report 14: Preliminary results of Ex2 presented in Carp., Lacid., Nic.

(2009) paper using


DEM.

Introduction

In the present report we shown preliminary results obtained simulating a compression test in a prismatic body test of
damaged concrete. The geometric and mechanic data of this test were obtained in the second example of the ( Fractal
analysis for damage detected in concrete structural elements under loading by Carpinteri, Lacidogna &Nicolini, Chaos
Solitons and Fractals 42(2009)2047-2056).)

The characteristic of the test

In the first four figures were extracted of the cited paper. These Figures depict the geometric and mechanic
characteristics of this example, and present the main experimental results.

Other important information extracted of the paper is cited at follow:


The prescribed velocity in the experimental test was (dδ/dt= 1.0 e-5mm/sec) from 0.0 to
1.0 e4 sec, and (dδ/dt= 1.0 e-5mm/sec) from 0.0 to 1.0 e4 sec.

Using this information about the test prescribed velocity and the Figure 2. it was possible obtained the initial elastic
modulus ( E=9Gpa).

This value of E is lower that classical concrete value for E. It is due to the fact that the body test before the test was
submitted to the pre-damage process ( The body was submitted to uniform load compression of 1300 KN during two
days then the body was unloaded).
At follow this damaged body was reloading and in this second moment the acoustic emission sensor were activated .
In Figure 1 the position of the sensors and the dimension of the body test is presented. In Figure 2 the results in terms of
reaction force vs time and a view of the layout of the experiment. In the Figure 3 we present the results in terms of b
coefficient for the three regions indicated in Figure 2, and finally in Figure 4 the final configuration obtained in the
experimental test is presented.
Figure 1 The body test geometry where were indicated the position of the EA sensors

Figure 2 : The results in terms of Load vs time and a view of the layout experiment.
Figure 3: Results obtained from experimental data in terms of b coefficient.

Figure 4: The final configuration obtained in the test.

The preliminary input parameters used in the DEM simulation were

E=9Gpa
Rho= 2500kg/m3 ( this value is common in concrete)
Gf(mean value)=500N/m ( with this value we consider all the energy that could be dissipated in the process. )
εp(mean value)=2.3e-3 ( critical strain).

With this value the program compute the following internal parameters (Kr=1.22, Rfc=9.5). In the Figure 5 a elemental
law of the DEM is presented.
a) b)
F F
Gf . Af
Pcr Lc Pcr
1 1
EA EA
kr - 1

εp εr = kr . εp ε εp εr = kr . εp ε

Figure 5: the elemental constitutive law used in the simulations.

Also was defined a variability in the field of Gf using a Weibull distribution and a correlation length equal to the level
of discretization.
The characteristic of the random present as result in terms of variation coefficient of CV(Gf)=50%, and CV(εp)=25%.
In the Figure 6 we present a diagram where is shown how the elemental constitutive law change among the bars of the
model.

F
Pcr

εp ε

Figure 6: The elemental law and its variation when we consider a Gf as random field.

The sensors were localized in the position indicated in Figure 1 ( SA1,SA6). And the lateral acceleration of the sensors
SA1-SA6 were captured during the virtual test.

A low viscosity damping was introduced in the model to kill the most high frequencies.
The test was carried out using a fast prescribed velocity taking care in not introducing important inertial effect during
the test as it is possible verified in Figure 8.

Preliminary Numerical Results

At follow we present the results obtained in the simulation, in Figure 7 we present the results in terms of reaction load
vs strain. We present in the same figure the experimental results. It is possible to see that it is necessary adjust better the
input parameters. But with this results let us shown the possibilities of this kind of simulation. In this moment I am
working to carried out this fine calibration of this model.
Experimental Result

Numerical Sim.

Figure 7 : Comparison of experimental and numerical results in terms of global stress and strain.

In Figure 8 we present numerical results in terms of energy balance during the numerical test.

Figure 8: Results in terms of energy balance ( the external work it wasn’t plotted).

In the Figure 9 results in terms of virtual final configuration are also presented. In this figure is presented a position of
nodes in a internal slice of the numerical model. It is possible see a clear localization in this final numerical
configuration. In the same figure is also presented the final configuration obtained in the experimental test.
Figure 9: Results in terms of final configuration, comparison between numerical and experimental results.

In Figure 10 we present the results in term of the acceleration measured in the SAE1 position ( see Figure 1).
Notice that the value of the aceleration it is to high, because they are obtained for the numercial twice derivation of the
displacement obtained during the simulation using a very simple algorithm.

Figure 10 : Results in terms of lateral aceleration measured in the SAE1 position ( see Figure1).

In Figure 11 we present a detail of the record where two kind of peak were indicated, the high peaks, indicated with
red arrow, and a precursor peaks indicated with black arrow. We called precursor peak in this context the little peak that
appear immediately before the big peak. In the computation of the peak to compute the b coeficient we consider both
two kind of peaks and as it is posible see in the Figures 16 and 17, this produce as result two declivity in the diagram.
(I think that this is a point to discuss).
In Figure 12 we present the region where up to me it is posible identified peaks. A new simulation with slow load we
are carrying out now. (I have doubt about the criterion adopted).

Figure 11: Detail in the end of the profile presented in Figure 10. With red arrow we mark the high amplitude peaks,
with black arrow we mark the precursors peak.

Figure 12 : Detail in the profile presented in Figure 10. We mark with black arrows where we begin consider peak in
our analisis.
Figure 13: Detail in the profile where appear all the sensor results. It is posible appreciate wich sensors are activated
first. See the position of the sensor in Figure 1. The prescribed velocity was imposed in the superior part of the body
test.

At follow in the Figure 14 we present a graph that show several results obtained in function of time.
This figure show that probably will be posible use the increment of damaged energy or increment of kinetic energy to
determine a b coeficient. ( We are working in this issue now).

Nacum=78

Damaged E.

Stress

Ninst=8

Kinetic E.

Figure 14 : Differents results parametes vs time during the simulation process.

In the Figure 14 bis we presented the Damaged Energy, Kinetic Energy and SAE1 aceleration vs time.
In this Figure it is posible to apreciate that in the present case there are a good correlation between the magnitude of the
aceleration peak and the abrupted steps in the Damaged Energy record . But in examples with complex geometry or
boundary conditions this correlation could to fail, that is, some damage could not be captured by determined sensor.
(This topic is touched in recent work of Rinaldi and I think that we could explore with DEM).
10xKinetic E.

SAE1 sensor x10-4

Damaged E.

t[s]

Figure 14 bis: Values of Damaged Energy, inetic Energy and Aceleration in sensor SAE1 vs Time.

In the Figures (15,16,17) we present the b coeficients computed using the SAE1 lateral aceleration.
The region to compute the values of b were divided in three region, that is: [ 0 to 0.5 sec], [0.5 to 1.0 sec.], and [1.0 to
tfinal]. We haven’t adopted a crietrium to identified the three region (other point to be discuss). Notice that the value
of b decrease when the damage increased showing that in this case the localization increase (b[0,0.5s]=1.82, b[0.5,1s]=1.67,
b[1.0,tfinal]=1.15.) This tendency is in acording to the experimental results.
We also notice in the second and third region the presence of precursors of the great peaks resulting in the log Nacum
vs Log Ac diagram two declivities. We mark with a read line in the Figures the results due to precursor peaks.
Values between [0,0.5s]

1.6

1.4

1.2

1
log N

0.8

0.6

0.4

y = -1.8211x + 8.6198
0.2

0
3.80E+00 3.90E+00 4.00E+00 4.10E+00 4.20E+00 4.30E+00 4.40E+00 4.50E+00 4.60E+00 4.70E+00
log Ac

Figure 15 : Results in terms of Log Nacum vs Log Ac . b[0,0.5s]=1.82.


Values between [0.5,1.0s]

1.6

1.4

1.2

1
log N

0.8

0.6

0.4

0.2

0
0.00E+00 1.00E+00 2.00E+00 3.00E+00 4.00E+00 5.00E+00 6.00E+00
log Ac

Figure 16 : Result in terms of log Nacum vs log Ac . b [0.5,1.0]=1.67. Notice two declivity the marked with red line is
due to the precursor peaks.
Values between [1.0, tfinal s]

1.4

1.2

0.8
Log N

0.6

0.4

0.2

0
0 1 2 3 4 5 6
log Ac

Figure 17: Result in terms of Log N acum vs Log Ac. b[1.0, tfinal s]=1.15. Notice a second declivity indicated with a
red line.

Finally in the Figure 18 we shown the SAE1 record process with a window fourier transform.
This figure shows that the peak appear clearly in the graphics. perhaps using this informations it is posible perform a
criteria to identify the peaks when these aren’t still visibles in the time domain. In the Figure 18 it is posible identified
the peak before 0.20 sec, but as is shown in Figure 12 it is posible detect peak only after 0.3sec.
Figure 18: The SAE1 aceleration record process with window fourier transform.

You might also like