BEHAVIOUR OF STRUCTURES MADE WITH RECYCLED COARSE ... · behaviour of structures made with recycled coarse aggregates originated from the precast industry joÃo nuno noronha ramos

BEHAVIOUR OF STRUCTURES MADE WITH

RECYCLED COARSE AGGREGATES ORIGINATED

FROM THE PRECAST INDUSTRY

JOÃO NUNO NORONHA RAMOS VIGÁRIO PACHECO

Extended Abstract

Master’s Degree in Civil Engineering

Junho de 2014

June of 2014

Supervisors:

Prof. Dr. João Paulo Janeiro Gomes Ferreira

Prof. Dr. Jorge Manuel Caliço Lopes de Brito

1

1. Introduction and research significance

The use of recycled aggregates (RA) in concrete is a step towards sustainability, enabling a re-

duction of the use of natural resources and of the waste produced.

Attention paid to this kind of material increased after WWII, given the urgent infrastructural and

housing needs, combined with the quantity of debris generated in some cities damaged by air

raids. Hansen (1992) provided a state-of-the-art on RA at the end of the 20th century. Presently,

the knowledge regarding this material is much more substantial, mostly because of the interest

in the behaviour of recycled aggregate concrete (RAC) increased, due to social, political and

cultural aspects regarding environment protection.

It is, however, necessary to further investigate the impact of this kind of aggregate on the prop-

erties of actual concrete structures, since most of the studies made focus on the material (me-

chanical and durability-related) properties of RAC, stopping short of its structural performance.

In this context, the BARPINP project, partly described in this document, intended to evaluate not

only the mechanical and durability properties of concrete mixes produced with recycled coarse

aggregates, but also the structural properties of structures made with those mixes. Since the recy-

cled aggregates were sourced from elements of the precasting industry, a high quality material was

obtained and a reduced influence of RA incorporation was expected. This document is related

only to the experiments regarding the structural behaviour of the different concrete mixes.

The innovative aspects of this experiment concern the use of RA from the precasting industry, and

the fact that, to the author’s best knowledge, this is the first study ever regarding the dynamic

behaviour of full-scale recycled concrete structures, the behaviour of full-scale recycled con-

crete slabs, the behaviour of a full-scale recycled aggregate concrete structure subjected to hori-

zontal loads and the first study regarding the behaviour of three-dimensional recycled aggregate

concrete frames, subjected to monotonic loading. It is also relevant that, contrary to most experi-

ments, these structures were executed in a common construction environment, with some choices

made to comply with such working conditions (like a simpler aggregate size grading). Thus it can be

stated that this study aims at replicating the performance of RAC structures as if they were built for

common construction purposes, and not for research purposes.

2. Objectives

This study aims not only at assessing whether the performance of RAC structures follows the

assumptions made when designing conventional reinforced concrete structures, but also at esti-

mating some core parameters of the structural behaviour, such as elastic modulus, natural frequen-

cies, cracking and ductility. To fulfil this purpose four full-scale recycled concrete structures,

made with various recycled aggregate ratios, were tested. Herein is described the design and mod-

elling of the test structures, as well as various tests and analysis of their results, namely dynamic

characterization tests, vertical loading tests and horizontal destructive tests.

The dynamic characterization tests allowed the study of the natural frequencies and dynamic be-

havioural pattern of the concrete compositions, as well as the estimation of the global elastic mod-

ulus of each concrete composition, by finite element modelling. The vertical loading tests intended

to study the execution conditions and stiffness of the slabs, with data concerning the midspan de-

flections of beams and slabs; these tests only studied the elastic response of the structures, to avoid

different levels of damage on the structures, which would influence the results of the last studies

2

carried out, by means of horizontal destructive actions, that replicated a pushover load, as defined

in Eurocode 8. These horizontal destructive tests allowed the comparison between force / defor-

mation curves, the evaluation of the ductility and cracking behaviour, the overall pattern of behav-

iour, hinge formation and the collapse mechanism of the test structures.

3. Experimental programme

Materials

Natural aggregates (NA) and recycled coarse aggregates (RA) were used. The NA used were lime-

stone coarse gravel and river sand. The recycled aggregates came from concrete blocks used in

the precast industry as support for long-span beams, with compressive strength well above 50

MPa. These concrete blocks were processed in a crushing plant, originating aggregates with a proper

grading (maximum aggregate dimension of 20 mm).

Composition of the concrete mixes

The objective was to develop concrete mixes within Eurocode 2’s class C25/30. Prior to defining

the four mixes of the structures, several concrete mixes produced in laboratory conditions were

tested for compressive strength, with the aim of determining a mix with the highest possible RA

content but without decreasing significantly the resulting concrete’s strength, and a mix with a

feasible content of superplasticizer (SP) that maximized the gains in compressive strength, for

total replacement of the coarse fraction of the aggregates. The other two mixes were a reference

mix (REF) and a mix with no SP and in which all coarse aggregates are RA (B100).

Certainly due to the high quality of the RA coming from precast elements, it was found that in

fact concrete strength slightly increased with the ratio of RA. Hence the four mixes used in this

campaign were as follows: REF - reference mix; B25 - a mix with the maximum ratio (25%) of

RA allowed by the Portuguese Laboratory of Civil Engineering (2006) and other institutions

(Gonçalves and de Brito, 2010), with the intent of replicating the properties of a conventional

concrete; B100; and B100SP - a mix similar to the previous one but with 1% superplasticizer by

cement weight. Table 1 contains the composition of each of these four mixes.

Table 1 - Concrete mixes’ composition (kg/m3 of concrete)

Concrete mix

Material REF R25 R100 R100SP

Fine sand 249.62 243.78 245.05 258.84

Coarse sand 543.3 448.79 450.12 475.45

Coarse natural aggregates 4-11.2 mm 358.42 328.05 0 0

11.2-22.4 mm 645.75 500.1 0 0

Coarse recycled aggregates 4-10 mm 0 83.32 333.27 352.02

10-20 mm 0 170.96 683.83 722.31

CEM II A-L 42.5R cement 350 350 350 350

Tap water 185.5 185.5 185.5 143.5

Superplasticizer 0 0 0 3.5

The aggregate’s grading followed a request from Opway (the construction company that supported

this project), based on practicability on site - the difference in maximum size between the NA and

the RA was due to the crushing process and the impracticality of sieving the natural aggregates.

The reinforcement steel used was A500, with B class of ductility. Table 2 shows the results of

the tensile strength tests, for rebars taken from the same shipment as the one used on-site.

3

Table 2 - Results of the tensile strength tests done on rebars

Average values fyk εyk ftk εuk k

533.8 MPa 2.0% 622.3 MPa 13.7% 1.17

This experiment only considers the use of the coarse fraction of the recycled aggregates, since the

fine fraction tends to have a higher percentage of attached mortar, thus resulting in a larger loss of

durability and mechanical properties (González-Fonteboa and Martínez, 2007; Sato et al., 2007).

Regulations and design of the test-structures

The design of the test-structures complied with Eurocode 2 (CEN; 2004a), Eurocode 7 (CEN;

2004b) and Eurocode 8 (CEN; 2004c). Due to testing limitations regarding the horizontal load

tests, concerning not only the maximum capacity of the grip hoists used in the load application

process, but also the costs of the foundations and anchor block, an adaptation was made: the

longitudinal reinforcement of the columns had to be reduced and did not fully comply with Eu-

rocode 8’s requirements regarding the formation of plastic hinges in the beams in column-beam

joints. Despite this, it was experimentally verified that, in the first floor, the hinges occurred

first in the beams, as initially intended. The design stage had to guarantee the following condi-

tions, regarding the horizontal load tests and respective structural behaviour: a ductile behaviour

(characterized by the formation of flexure hinges), the technical and economic feasibility of the

experiment and a simple test setup that would guarantee the operation of the equipment without

errors.

The computer program used was CSI’s SAP2000 (v15) and the loads were applied taking into

consideration the geometry of the structure and an inverted triangular horizontal distribution,

with loads applied to both floors at the slab level. Table 3 shows such forces, applied in nodes

as a “displacement controlled load case”. This kind of load is defined as a vector (last column of

the table), and for each step the software applies a factor that increases or decreases the forces

(during loading or unloading stages) in comparison with the previous load step.

Table 3 - Loads setup

Ratio between horizontal forces in the upper cables and lower cables: 2.05

Cable Vector direction Length %/direction Normalized force

Cable 1

Pushover 17.05 m 95.36% 205.00

Orthogonal/horizontal 1.40 m 7.83% 16.83

Orthogonal/vertical 5.20 m 29.08% 62.52

Cable 2

Pushover 18.76 m 96.12% 205.00



Cables 3 and 4

Pushover 18.00 m 99.07% 100.00



fy - yielding stress

ft - maximum stress

k - hardening factor

εyk - yielding strain

εuk - strain at maximum

stress

4

For the purpose of designing the test set-up, the finite element model (FEM), used for both the struc-

tures and the anchor block (as well as the rest of the load application system), was made considering

the concrete’s (class C25/30) 95% characteristic material properties, instead of the normally used 5%

characteristic values - the material characteristics of the reinforcement steel were, at this stage, esti-

mated considering that the yielding stress was 1.3 times the characteristic design value (the maxi-

mum capacity specified by Eurocode 2). The nonlinearities were modelled as fibre plastic hinges (at

the ends of the columns and beams) and the model suggested by Mander et al. (1988) was used to

estimate the properties of confined concrete. The maximum forces of the cables during the experi-

ment, predicted by the FEM, were used to design the anchor block; Eurocode 7 was taken into con-

sideration, and possible undesired collapse mechanisms (such as shear failure of the structural ele-

ments or toppling) were accounted for, considering a safety factor of 1.5. The anchor block was 1.7

m high, 1.2 m of which sunken into the ground (mobilizing passive impulse) and has a square shape

with 2.75 m sides. A grid of reinforcement bars was used, as well as a system with 1 m screws and

threaded couplers, which allowed the attachment of the bi-articulated anchor points.

Model’s characterization

Figure 1 shows the models geometry and reinforcement layout. The slabs are 0.10 m thick and

their reinforcement is a bottom mesh of 8 mm rebars spaced 20 cm in both directions. Since the

foundations (Figure 2) are two large concrete blocks, a full base restraint of the columns is en-

sured (as confirmed during the horizontal load tests by the displacement values of topographic

targets 5 and 6, shown in Figure 13). The rebars cover in the columns is 2.5 cm and in the rest

of the structure 2.0 cm.

Figure 1 - Geometry and reinforcement layout of the test-structures

Since the size of the elements that make the structure is relatively small, deviations in geometry

can be meaningful; hence a geometry survey of the structure was performed, for FEM purposes.

The slabs’ thickness was considered equal to the length of concrete cores taken from their cen-

tres whilst the columns’ dimensions were considered equal to the average of two measurements

made along their length.

The slabs’ thickness ranged between 10.5 cm and 11.6 cm and the columns cross-section sides

varied between 19.4 cm and 21.8 cm. The density of each concrete mix was considered equal to

the average of those of 16 standard test cubes, made during the execution of each of the structures.

A FEM was made for each structure, considering these dimensions and densities.

5

Figure 3 - Location of the transducers

Figure 2 - Geometry and reinforcement layout of

the foundations

4. Experimental procedure

4.1. Dynamic characterization tests

Definition of the test setup and test procedure

The four structures were tested 65 ± 5 days after the execution of the second floor.

All the tests were performed twice: once with two piezoelectric unidirectional accelerometers,

connected to charge amplifiers and a data acquisition unit and a second time with a triaxial strong-

motion seismograph. For the horizontal vibration tests the accelerometers were placed in the col-

umn/beam joints, in the direction of the excitation, at each slab level separately during the hori-

zontal testing, whilst the seismograph was placed in two distinct locations in each floor: one at the

centre of the slabs, minimizing the torsional frequency readings, and another at the edge of the

slabs, with maximum eccentricity, maximizing those readings. For the vertical excitations, the

seismographs had the same location and the accelerometers were placed at the centre of each slab.

The location of the measuring systems is shown in Figure 3.

The characteristics of the readings and subsequent frequency analysis are given in Table 4. The

frequency ranges studied are not limited by the Nyquist frequency (the maximum frequency that

can be analyzed by a given test record, equal to half the sampling rate), due to the low-pass filters

applied on both setups. The frequency analysis was made using a FFT transform with 4096

(212

) data points.

Table 4 - Characteristics of the data acquired and of the frequency functions

Equipment Sample rate Nyquist frequency Low-pass filter

Accelerometers 300 Hz 150 Hz 100 Hz

Seismograph 200 Hz 100 Hz 80 Hz

Three kinds of excitation were studied: centered, eccentric and vertical (by jumping on top of the

slabs). The centered and eccentric excitations were imposed by manually pushing a rope, with

6

unspecified intensity. Figure 4 clarifies the horizontal excitations in the first floor of one structure

- similar excitations also occurred in the second floor of each structure. Additionally, the seismo-

graph was also used to measure ambient vibrations, for each test setup. The ambient sources of

excitation were mainly the wind and agricultural vehicles, functioning in a field nearby.

Figure 4 - Horizontal excitations performed (first floor)

Results and discussion

Vibration modes, frequencies and FEM calibration

The horizontal dynamic response of the four structures followed the non-calibrated FEM, both

in terms of vibration modes and frequencies. This numerical model predicted that the first and

third horizontal modal shapes were pure translations, whilst the third had a torsional configura-

tion. Figure 5 is an example (forced vibration readings done with the seismograph) of the fre-

quency functions obtained with this analysis.

Figure 5 - Average horizontal frequencies and confidence intervals (forced vibrations, seismograph)

The elastic modulus (E) calibration analysis (discussed below) provides a better comparison

between structures, namely due to geometry differences between them. For instance, the higher

frequencies of R25 comparing to REF’s could be a consequence of thicker columns (hence

higher transverse stiffness), instead of a higher E.

All the tests were used to estimate the first horizontal mode, whilst for the second mode the seis-

mograph’s readings used were the ones with the seismograph in the eccentric position and with

the structure subjected to torsional excitation. The accelerometers’ results used for the second

mode were the ones with the eccentric excitation, considering the difference between the accelera-

tions measured in each accelerometer (for the first and third mode the value used was the average

of the readings). The third mode was considered in all the tests performed, when significant.

7

The results of the ambient and forced vibration tests, as well as of readings made with the seis-

mograph and with the accelerometers, are consistent.

The results suggest a decrease in frequency due to the incorporation of RA, even though R25

presents slightly higher frequencies than REF, probably because of variations in the curing and

execution conditions (in Figure 5 there is an overlap in confidence intervals). However, the dif-

ference between both R25 and REF when compared to R100 strongly suggests a decrease in

frequencies with AR incorporation, following the studies regarding E and deformations present-

ed in the literature review. R100SP showed the highest frequencies, due to the increase of E

caused by the use of SP, as obtained in the laboratory tests of the small specimens and as Bar-

budo et al. (2013), Matias et al. (2013) and Pereira et al. (2012) argue.

The vertical frequencies are highly dependent on the thickness of the slabs, as seen in Table 5,

where no RA-trend can be perceived. The vertical frequencies’ results have a higher scatter than

the horizontal ones, mostly due to the lower number of samples. The confidence intervals were

determined for a significance level of 0.05.

Table 5 - Statistical parameters of the vertical frequencies’ analysis

REF R25 R100 R100SP

Slab Slab Slab Slab

Equipment Excitation Parameters 1st 2nd 1st 2nd 1st 2nd 1st 2nd

Seismograph

Ambient

Samples 2 2 2 2 2 1 2 2

Average (Hz) 32.75 26.1 37.27 35.85 32.85 30.64 36.48 38.17

Standard

dev. (Hz) 1.06 0.48 5.04 3.32 1.35 - 3.73 0.38

Confidence

(Hz) 1.47 0.67 6.99 4.60 1.87 - 5.17 0.53

Forced

Samples 4 4 4 4 4 4 4 4

Average (Hz) 32.15 24.36 33.97 30.54 31.00 30.98 38.63 38.49

Median (Hz) 32.38 24.35 34.35 30.99 31.04 30.97 38.54 38.44

Standard

dev. (Hz) 0.50 0.46 1.99 1.06 0.28 0.14 0.17 0.17

Confidence

(Hz) 0.49 0.46 1.95 1.04 0.28 0.14 0.17 0.17

Accelerometers Forced Samples 2 2 2 2 2 2 2 2

Average (Hz) 32.0 25.2 32.9 31.1 31.4 31.1 37.0 37.6

The Young’s modulus’ calibration was done individually, for each test, ensuring a richer statis-

tical analysis. A higher emphasis is made concerning E values than frequency ones, given that

the different geometry of the structures partly distorts a frequency analysis. The criterion was to

calibrate E value in a way that the first numerical horizontal frequency would be the same as the

experimental. The evident shrinkage of the upper slab of REF was taken into account, by reduc-

ing the E value of that slab to 65% of the value of the remaining structure - this value was sug-

gested by the deflections of the vertical static loading test, discussed further on. Tables 6 and 7

show, for each measuring system, the average values of E after calibration and the relative er-

rors for each vibration mode (higher than the first) and measuring system.

Figure 6 shows the results of E calibration of the forced vibration tests made with the seismograph.

To better evaluate the experimental significance of these results, t-tests for unequal variances

were made with the values of each calibrated E, comparing the three structures with RA with

the one with REF: the results showed that on one hand R100SP’s elastic modulus is statistically

higher than REF’s, whilst R100’s modulus is lower than REF’s. The comparison between R25

and REF was statistically inconclusive. The significance level used for the t-test calculations

was 0.05, the same as that used in the calculation of the confidence intervals. A final compari-

son between these results and the laboratory ones is made in Table 8, which shows that the dif-

8

ference in elastic moduli between the environments is low (below 5%).

Table 6 - E and relative error (experimental / numerical frequencies) after calibration (seismograph)

Relative error

(numerical and experimental frequencies)

E modulus Horizontal modes Vertical modes

Structure Parameter (GPa) 2nd

mode 3rd

mode 2nd

slab 1st slab

REF

Samples 16 4 14 4 4

Average 35.31 4.05% 0.97% 9.75% 4.19%

Median 35.10 5.21% 0.89% 9.80% 3.51%

Standard dev. 0.60 2.33 0.85 1.72 1.49

Confidence 0.29 2.28 0.45 1.68 1.46

R25

Samples 16 4 12 4 4

Average 35.82 6.69% 2.10% 3.10% 4.51%

Median 36.13 6.64% 1.86% 3.05% 4.17%

Standard dev. 1.72 1.43 0.65 0.89 2.69

Confidence 0.84 1.40 0.37 0.87 2.64

R100

Samples 15 6 14 4 4

Average 30.72 7.09% 7.83% 9.82% 2.42%

Median 31.17 7.12% 8.29% 9.78% 2.30%

Standard dev. 0.69 0.90 1.37 0.50 0.88

Confidence 0.35 0.72 0.72 0.49 0.87

R100SP

Samples 16 4 16 4 4

Average 41.03 4.75% 2.38% 17.27% 5.02%

Median 41.06 4.34% 2.39% 17.13% 4.79%

Standard dev. 0.33 1.15 0.45 0.53 0.46

Confidence 0.16 1.13 0.22 0.52 0.45

Table 7 - E and relative error (experimental / numerical frequencies) after calibration (accelerometers)

Relative error

(numerical and experimental frequencies)

E modulus Horizontal modes Vertical modes

Structure Parameter (GPa) 2nd

mode 3rd

mode 2nd

slab 1st slab

REF

Samples 4 4 4 2 2

Average 33.99 5.83% 0.96% 13.93% 3.34%

Median 33.99 5.83% 0.59% 13.93% 3.34%

Standard dev. 0.36 1.39 1.08 2.29 1.56

Confidence 0.35 1.37 1.06 3.18 2.17

R25

Samples 4 4 4 2 2

Average 36.15 3.37% 3.68% 2.36% 5.45%

Median 35.52 3.17% 3.66% 2.36% 5.45%

Standard dev. 1.79 1.44 0.04 0.81 5.69

Confidence 1.76 1.41 0.04 1.13 7.88

R100

Samples 4 4 4 2 2

Average 30.66 6.18% 8.38% 10.24% 10.24%

Median 30.32 6.35% 8.31% 10.24% 10.24%

Standard dev. 0.68 1.12 1.09 0.37 0.37

Confidence 0.66 1.10 1.06 0.52 0.52

R100SP

Samples 4 4 4 2 2

Average 33.99 6.26% 2.83% 14.60% 3.57%

Median 33.99 6.44% 3.01% 14.60% 3.57%

Standard dev. 0.36 1.96 0.72 4.31 0.88

Confidence 0.35 1.92 0.71 5.97 1.22

9

Figure 6 - Average E modulus after calibration and confidence intervals (seismograph)

Table 8 - Comparison between calibrated and laboratory elastic moduli

Structure

Origin of the results REF R25 R100 R100SP

Laboratory 36.20 34.10 32.10 39.50

Ambient vibrations Calibration average 35.31 33.37 31.17 40.79

Ratio with laboratory 0.98 0.98 0.97 1.03

Forced, seismograph Calibration average 35.37 35.82 30.72 41.03


Forced, accelerometers Calibration average 34.73 35.63 30.83 40.13


Some observations should be made concerning these results:

A high relative error is observed in the second floor slab of the REF structure (already con-

sidering the difference in stiffness according to the results of the vertical static testing);

R100SP’s relative error in the second floor slab is the highest, probably due to execu-

tion conditions (SP concrete is more susceptible to differences in mixing);

A lower relative error was obtained in the vertical modes of the other slabs when the

calibration process included a calibration specifically done for their material properties

(such consideration was not taken into account since it could provide artificially accu-

rate results, with no physical validity);

There is an overlapping in confidence intervals between REF’s and R25’s elastic modu-

li and, most probably, the reason for the higher average stiffness of R25 is the different

experimental and execution conditions. The t-test comparison between these structures

showed that their difference is not statistically conclusive;

R100SP’s mix proved that the use of SP increases the stiffness of RAC mixes;

The good quality of the aggregates generated from precast elements was proven, with a

decrease in E of about 10 %, relative to a conventional concrete mix, for total replace-

ment of the coarse fraction of the aggregates;

The trend (and results) of E calibration for the various structures follow laboratory re-

sults, with ratios between them always lower than 5%.

Modal damping

The modal damping (ξ) was estimated using the logarithmic decrement of the forced vibration tests.

Since there was some superposition of frequencies, a signal filter was applied and only the first hori-

zontal vibration mode of the structures and the first vertical vibration mode of each slab were stud-

ied. The calculations were made for four cycles (roughly one second for the horizontal modes), the

decrement technique was used for both the maximum and the minimum of each cycle and ξ was

considered as the average of these values. Figure 7 clearly shows a significant scatter, which was

expected, not only because this is a frequent problem of experimental modal damping estimation

(Magalhães et al., 2010), but also because the excitations were manually controlled, with different

amplitude and frequency for each test done.

10

Figure 7 - Average damping estimation and confidence intervals

Additionally, by combining the results of the tests of different excitations (torsional or centered, rope

attached on the first or second floor, …) different damping mechanisms occur. Nevertheless, it was

chosen to combine the results, since the different excitations were done in all structures, averaging

this effect and the effect of different excitation frequencies and amplitudes of the various tests.

This significant scatter of results and the erratic averages do not allow conclusions on the effect of

AR on the structural damping of RAC structures. This is an area of study to be addressed in the fu-

ture, since no related studies were found by the author.

4.2. Vertical load tests


The four structures were tested at an age of about 120 days, with a 20-day margin.

Each slab was tested individually, with the following equipment: a) displacement transducers

(placed on the midspan of each beam and at the centre of the slabs); b) a datalogger, to record

the displacements; c) a hose to enable filling the floor with water; d) four rulers.

The four rulers allow a continuous monitoring of the water height, enabling the registry of the

uniformly distributed load applied on the slab at any given moment; the maximum height of water

was defined as 50 cm, corresponding to a uniformly distributed load of 4.9 kN/m2. This is a typi-

cal serviceability limit-state value for a slab load considering the sum of dead and live loads. The

non-elastic behaviour of the slabs was not analysed since further testing (a destructive test by

means of horizontal loads) was planned, and it was fundamental to prevent potential differences in

behaviour between structures, caused by different cracking stages of slabs and beams.


Comparison of the deflections of the beams and slabs

Figure 8 shows the deflections obtained for the two slabs and corresponding support beams of

the B100SP structure, previous to data post-treatment. As expected, an elastic behaviour and

very small deflection values were witnessed.

Figure 9 shows the average deflection values of the beams of each slab, whilst Figure 10 shows

the slabs’ deflection after subtracting the average values of the corresponding beams of the slab.

These values are herein named “relative deflections of the slabs”.

Since the behaviour was elastic, and to correct small oscillations on the test setup due to strong

wind (relevant since the measured deflections are very small), the deflection values shown in

both figures correspond to the average of the loading and unloading stages.

11

First floor Second floor

Figure 8 - Load-displacement curves of the B100SP structure


Figure 9 - Load-displacement curves of the beams of the four structures

Deflection results


Normalization taking into account the slabs’ thickness


Figure 10 - Load-displacement curves of the slabs of the four structures

Figure 9 shows that the deflections of the beams are unaffected by the concrete mix. This is ex-

0

1

1

2

2

3

3

4

4

5

5

0,00 0,10 0,20 0,30

Load

(kN

/m2)

Midspan deflection (mm)

Slab

Beam 1

Beam 2

Beam 3

0

1

2

3

4

5

0,00 0,10 0,20 0,30

Load

(kN

/m2 )


Slab

Beam 1

Beam 2

Beam 3

Beam 4

0

1

2

3

4

5

0,00 0,04 0,08 0,12

Load

(kN

/m2 )

Midpspan deflection (mm)

REF

B25

B100

B100SP

0

1

2

3

4

5

0,00 0,04 0,08 0,12

Load

(kN

/m2 )


REF

B25

B100

B100SP

0

1

2

3

4

5

0,00 0,05 0,10 0,15 0,20 0,25 0,30

Load

(kN

/m2 )


REF

B25

B100

B100SP

0

1

2

3

4

5

0,00 0,05 0,10 0,15 0,20 0,25 0,30

Load

(kN

/m2 )


REF

B25

B100

B100SP

0,0

1,0

2,0

3,0

4,0

5,0

0 100 000 200 000 300 000

Load

(kN

/m2 )

Midspan deflection x slab thickness3 (mm4)

REF

B25

B100

B100SP0,0

1,0

2,0

3,0

4,0

5,0

0 100 000 200 000 300 000

Load

(kN

/m2 )

Midpsan deflection x slab thickness3 (mm4)

REF

B25

B100

B100SP

12

plained by the masonry walls (height of about 65 cm) that increase the stiffness of the beams to

such an extent that differences in the elastic modulus of the concrete mixes become almost irrele-

vant. This is evidence that, as confirmed by the laboratory and the dynamic analysis tests, the dif-

ferences in elastic modulus between mixes are not significant from a structural perspective.

Figure 10 also shows the relative deflection of slabs and the relative deflection multiplied by the

cube of the slab thickness (the deflection depends linearly on this parameter). This normaliza-

tion relative to the thickness of each slab allows a direct comparison between the slabs’ material

properties, since the other parameters involved (load and span) are the same for the four struc-

tures. This figure also emphasizes REF’s second floor slab’s high deflection (hence lower stiff-

ness and modulus of elasticity of this concrete mix), contrary to what was shown in the litera-

ture review and observed in the laboratory and the dynamic tests. This deviation is most proba-

bly caused by shrinkage cracking, observed on-site and caused by inadequate curing and ex-

tremely hot weather during casting. Since the dynamic tests analysed primarily the horizontal

behaviour (mostly dependent not on the slabs but on the columns, subjected to lower shrinkage

effects due to lower exposure area and the presence of formwork during the first stages of cur-

ing) this effect was not noticeable.

Finite element models and modulus of elasticity of each structure

The finite element models had to take into consideration the following aspects:

The masonry walls increase the torsional stiffness of the slabs considerably, making the

slabs behave as fully supported;

The Young’s modulus of the masonry walls was set at 3 GPa after a preliminary analysis;

The need to consider a different equivalent elastic modulus (taking into account the effects

of cracking) for the second floor of the REF structure, given the relatively advanced

“cracking II stage” of the slab, which made its stiffness lower than the one of the slab in

the first floor and of the slabs in the other structures;

It was decided not to consider in the analysis the second floor slab of the B100SP struc-

ture, because its deflection was completely outside the expected range, due to experi-

mental difficulties that were not detected on-site.

Table 9 shows the modulus of elasticity values, relative errors and other data analysis after the finite

element modelling.

Table 9 - Results after finite element modelling

REF B25 B100 B100SP

1st floor

E (GPa) 35.4 39.5 34.9 44.8

Slab thickness (cm) 10.5 11.0 11.6 10.5

δ beam - exp.(mm) 0.083 0.084 0.082 0.086

δ beam - FEM (mm) 0.092 0.091 0.089 0.090

Relative error - beam

(%) 9.8% 8.1% 8.3% 4.4%

δ slab - rel. exp. (mm) 0.189 0.175 0.156 0.164

δ slab - rel. FEM(mm) 0.199 0.175 0.171 0.176

Relative error-slab (%) 5.0% 0.0% 8.8% 6.8%

2nd floor

E (GPa) 23.0 39.5 34.9 44.8

Slab thickness (cm) 10.5 11.2 11.6 10.7

δ beam - exp.(mm) 0.101 0.116 0.109 0.107

δ beam - FEM (mm) 0.101 0.096 0.095 0.094

Relative error - beam

(%) 0.3% 20.6% 15.2% 11.97%

δ slab - rel. exp. (mm) 0.278 0.154 0.173 -

δ slab - rel. FEM(mm) 0.267 0.174 0.175 0.174

Relative error-slab (%) 4.1% 11.5% 1.1% -

13

The second floor slab of the REF structure had an E value 35% lower than that of the first floor. A

comparison between the results obtained in the four structures and the ones obtained in laboratory

test specimens shows close results (differences of around 10%, explained by different execution

conditions). It is also evident that not only the second floor of REF suffered shrinkage cracking, but

also the first, to a lower extent than the second, since its calibrated equivalent elastic modulus is

lower than the one of the B25 structure (the laboratory results suggested otherwise). Furthermore,

there were no significant differences in deflections and E values between different concrete mixes.

4.3. Horizontal destructive load tests


The structures were tested individually, with a controlled load distribution.

After considering costs, manoeuvrability and portability of the equipment to be used in this exper-

iment, it was decided to apply forces on the structures using grip hoists, with their cables attached

to the column-beam joints as shown in Figure 11. In the same figure, most of the remaining ele-

ments of the test setup are shown, as well as the topographic targets used to measure displace-

ments. The four anchor points (only two are shown) were attached to a concrete anchor block,

represented in Figure 12.

Figure 11 - Overview of the test setup of one structure Figure 12 - Anchor block and test-structures

Due to the large scale of this project, it was necessary to design a test setup that would reutilize

materials and equipment, reducing costs and space required for the experiment. Hence, a cross-

shaped solution was chosen (the same concrete anchor block was used to test all the structures).

This solution only allows loading on one direction, emulating a pushover analysis, whose loads

are monotonic.

The use of grip hoists to apply the loads makes this experiment an hybrid between force controlled

tests and deformation controlled tests, as on the one hand the loads are applied taking into consid-

eration the measurements of the load cells (tension in each cable), but on the other hand the same

loads are caused by the strain on each cable, and this strain depends on the deformation of the

structure, decreasing with an increase in deformation.

This allowed the experiment to continue after the maximum load capacity was reached, enabling

the capture of part of the loading decrease (softening) of each structure.

The structures were tested in two consecutive days. A data logger and four tension load cells

were used to measure the loads applied by each cable, ensuring a load application in accordance

14

with the distribution shown in Table 3. A digital inclinometer was used to measure the angles of

the cables and of the load cells with the horizontal, to account for the loads effectively applied

in the direction of the pushover experiment (horizontal).

The load was applied in increments/decrements of 5 kN and at the end of each step the dis-

placement of the topographic targets was measured. When cracks started to appear in the lower

half of the two columns tensioned by the cables, crack width measurements were registered, as

well as the crack spacing at the end of the experiment. Most of the topographic targets placed at

the column-beam joints at the first floor level of the structures could not be used at the final

stages of the experiment, due to the loss of the concrete cover they were attached to. Loss of

concrete cover (unconfined concrete) on the bottom of the lower columns was also witnessed,

due to their low compressive strain capacity. Figure 13 shows some details of the experiment.

Figure 13 - Some operations during one test


Behaviour of the structures

The behaviour of the structures followed the same pattern, thus only the behaviour of one of them

(B100) will be described in detail. Since the purpose of the topographic targets 5 and 6 was to

confirm the total restraint of the foundations, as well as the absence of toppling, both confirmed,

no further mention of their measurements is made.

At this stage, the actual in situ properties of the reinforcement steel and each concrete composition

were known and the FEM was updated for each structure, considering its concrete composition.

Figure 14 shows the displacement at second floor level caused by the forces and corresponding mo-

ments (at ground level). This figure shows that the load distribution followed the targeted inverted

triangular load distribution shown in Table 3, since the bending moments are roughly five times

higher than the acting forces, as Equation 1 clarifies.

1 1 2 2

1 2

st st nd ndBase

st nd

Force floor x Height floor Force floor x Height floorMoment

Shear force Force floor Force floor

3 2 6 15 5

2 3

Base

Moment F x F x F

Shear force F x F F

(Eq. 1)

Cracking of the columns only started to be noticed after these elements reached relevant deflec-

tions (stage 1 in Figure 14). The yielding of the cross-sections of (future) plastic hinges at both

ends of the columns of the first floor (stage 2) was caused by basal forces of about 50 kN, corre-

sponding in the FEM to a bending moment at the ends of those columns of 20 kNm. The yield-

ing bending moment of those sections had been algebraically estimated at the cross-section level

as 20.8 kNm.

15

Figure 14- Experimental load/deformation curves - B100 structure

The maximum experimental loading capacity (stage 3) is 66 kN while the corresponding value in the

FEM is 63 kN. The 66 kN load is associated in the FEM with a bending moment at the bottom of the

two columns tensioned by the cables of 26.3 kNm, close to 24.8 kNm, the strength value algebraical-

ly estimated at the cross-section level, taking into consideration the hardening of the rebars and the

axial force on the column, 5 kN. The closeness between these values is an indication that the

measurements and calculations made were correct.

In terms of qualitative behaviour, high ductility was observed (Figure 15), mostly due to the

detailing design. This shows that RA use is by no means a hindrance to proper ductile behav-

iour, as long as conventional concrete code regulations are followed.

Figure 16 - Plastic hinge formation and collapse

mechanism

Figure 15 - Some details of the behaviour of B100 Figure 17 - Formation of plastic hinges in beams

The hinge formation and collapse mechanism are shown in Figure 16. After an initial behaviour

with no evident concentrated deformations in any cross section, the bottom of the ground-floor

columns starts to have a significant curvature and to exhibit cracks shortly afterwards; at this time,

16

and due to increased deflections, the hinges of the two beams of the first floor parallel to the push-

over direction become visually apparent and allow the beams and slabs of the first floor to remain

horizontal, as Figure 16 clearly shows. At this stage the column-beam joints of the first floor rotate

in agreement with the lower columns (as also shown in Figure 17), and the upper columns show

no evidence of plastic hinge formation but have a distributed curvature along their length, that

allows the upper beams and slab to remain horizontal with no hinge formation. Then, concentrated

deformations start to become visible at the upper ends of the columns of both floors, showing

plastic hinge behaviour of those sections. Collapse happens after the hinges of the first floor ex-

hibit significant rotations, clearly higher than the rotations of the plastic hinges of the upper end of

the columns of the second floor. The bottom of the columns of the second floor follows the rota-

tion of the corresponding column-beam joints and do not show a cracking pattern associated with

hinge formation. This is certainly due to the existence of plastic hinges at a high degradation state

at the corresponding top of the columns and beams of the first floor. During the last stages of the

test, concrete covers started to fail at the bottom of the columns due to lack of capacity of the

unconfined concrete to withstand compressive strains in the same way as confined concrete

does.

Comparison between structures

Figure 18 shows that the load vs. displacement curves of the four structures was equivalent,

showing no clear trends or noticeable influence of RA incorporation. The apparent higher max-

imum base shear of structures B100 and B100SP is caused by a slightly different force distribu-

tion than intended (higher tension on the first floor cables and lower tension on the second floor

cables). As the bending moment-displacement curve shows, the maximum moments are identi-

cal in all the test-structures.

Figure 18 - Comparison between force/deformation curves

These results agree with most of the structural studies of the literature review; for most reinforced

concrete structures, as long as proper design is performed, the load action-effects distribution de-

pends mostly on the steel reinforcement, independently of RA incorporation. This is easily ex-

plained by looking at typical equations of the yielding or maximum moment of cross-sections: the

dependence on the concrete properties is hardly significant. Thus, it would be expected that differ-

ences would only be found in the pre-cracking stage behaviour of the structures, with higher de-

flections associated with B100, due to its lower modulus of elasticity. However, most probably

due to shrinkage cracking of the REF structure (especially evident during the analysis of the verti-

cal load tests), which reduced its equivalent stiffness, REF’s deflections were the highest.

Nevertheless, the deflections of the four structures did not significantly differ and it can be conclud-

17

ed that the execution conditions are more relevant to the deformation behaviour of the structures

than the incorporation of these RA. The use of RA was already expected to have a negligible effect,

based on the results of other investigations (Xiao et al., 2006; Xiao et al., 2012; Wang e Xiao, 2013),

and also because the FEM, which consider the in situ properties of the various concrete mixes, led to

identical pushover results in all structures, namely in terms of shear capacity during the loading

stage, as Table 10 clearly demonstrates. The closeness between experimental and analytical base

shear results is evident in the same table. The unloading stage results differed from the numeric ex-

pectations, especially in displacements, due to the execution process of the load tests and to the non-

linear behaviour, not as accurately estimated as the linear behaviour, as also observed in Sharma et

al. (2013) which describes a similar experiment on a conventional reinforced concrete structure.

Table 10 - Base shear values associated with structural stages

Stage BR B25 B100 B100SP

Visible cracking Experimental 28.4 kN 27.5 kN 31.0 kN 28.2 kN

Numeric - - - -

Yielding Experimental 50 kN 52.8 kN 50 kN 67.7 kN*

Numeric 54.2 kN 55.0 kN 59.6 kN 62.3 kN

Maximum capacity Experimental 59.0 kN 62.5 kN 63.0 kN 67.7 kN

Numeric 63.2 kN 63.0 kN 63.6 kN 66.6 kN

The displacements and forces associated with the collapse of the structures are conditioned by the

way the experiment was executed, because it was not performed under controlled displacements

(an option that was not technically feasible). To unload the structures during the softening stage, it

was necessary to slightly load them (by tensioning the cables) at each equilibrium stage. The un-

loading was then made by the dissipation system structure/cables, since the cables’ push originates

displacements, reducing the load applied on the structures, which in turn decreases the strain in

each cable. During this process, if a grip hoist operator pushes the cable more quickly at a later

stage of a structure’s test it will collapse for a lower horizontal displacement than the other struc-

tures, due to excessive loading applied at that stage.

Excellent ductility was witnessed in all the structures (Table 11), which demonstrates that, if code

regulations are followed, RAC structures behave like conventional reinforced concrete structures.

Since the displacements at collapse depend on the test execution and no direct comparison can be

made with a common ductility parameter, Duct.u (ratio between collapse and yielding displace-

ments), an additional ductility criterion was used for comparison, Duct.0.5 (ratio between the sof-

tening displacement for half the maximum load capacity of the structure and the yielding dis-

placement). The Duct.0.5 values in Table 11 show that not only the ductility of the structures is

quite good, but also that RA has no significant effect on this parameter.

Table 11 - Ductility of each structure

Structure RC B25 B100 B100SP

δy 6.4 5.7 5.6 6.9

δu 66.0 47.4 48.0 58.8

Duct.u 10.3 8.3 8.6 8.5

δ0.5 57.0 46.1 46.3 52.6

Duct. 0.5 8.9 8.1 8.2 7.6

Cracking

Crack width and spacing were measured at the lower half of the two columns tensioned by the

cables. The load values associated with the different stages of cracking were also registered. For

comparison purposes, the crack width at each load stage was considered as the average of three

18

crack measures, whilst crack spacing was considered as the average of all the registered cracks of

both columns of each structure after the collapse (typically about 17 quasi-horizontal cracks).

As Figure 19 shows, structures REF, B25 and B100SP had a similar behaviour, which suggests

that RA incorporation did not influence the cracking behaviour of these structures.

Figure 19 - Crack width and spacing

Most of the studies in the literature review suggest that RA use results in higher crack widths and

spacing, which disagrees with this study’s findings.

This is explained by the influence of reinforcement steel in the cracking behaviour of concrete

structures, as seen by González-Fonteboa and Martínez (2007). Since the columns of the test

models have a low reinforcement spacing (as seen in Figure 1), it was expected that any differ-

ences in cracking behaviour due to RA would be significantly mitigated.

The crack spacing of the columns of the B100 structure is roughly 1.5 times that of the columns of

the other three structures. This was an unexpected result since the tensile strength of all concrete

mixes was very similar according to the laboratory tests and there were no significant differences in

terms of reinforcement cover.

5. CONCLUSIONS

These experiments aimed at studying the applicability of recycled coarse aggregates and assessing

the structural behaviour of structures designed in compliance with Eurocode regulations. Full-

scale models of three mixes including recycled aggregates, as well as a conventional concrete mix,

were produced and tested To the author’s best knowledge this is the first study regarding the be-

haviour of 100% replacement RAC structures subjected to vertical loads, the first study regarding

the dynamic properties of RAC structures and the first study regarding the behaviour of full-scale

recycled aggregate concrete structures subject to horizontal destructive tests.

By comparison between the properties the three different RA mixes and the reference mix of

conventional concrete, it is shown that there are no obstacles to the use of RA.

Furthermore, the behaviour of structures with this kind of aggregates is analogue to the well-known

behaviour of conventional concrete structures, the assessment of the dynamic properties is possible

with methods already in use and, for high quality recycled aggregates, the decrease in elastic modu-

lus is almost negligible, as shown both by the dynamic characterization and the vertical load tests -

nevertheless, the use of superplasticizers is a good solution to compensate the (low) loss in this prop-

erty. A proper seismic behaviour is viable, at least if code regulations are respected.

Particularizing for each individual test the conclusions drawn were:

19

Dynamic characterization tests

1. The behaviour pattern was the same for all structures, with natural frequencies and

modal configurations in agreement with the ones obtained in numeric models;

2. The values of the elastic moduli estimated with this method were consistent with the

ones obtained in the laboratory tests;

3. The use of RA without SP slightly decreases the Young’s modulus of concrete;

4. These results prove the applicability and good quality of these aggregates, generated from

precast elements, with a low reduction in elastic modulus (10%) and a decrease in mass

(4%) for a full replacement of the coarse fraction of the aggregates;

5. The use of SP in a mix with all natural coarse aggregates replaced by RA proved to be a

good option, and the resulting elastic modulus is the highest of the various mixes tested;

6. The modal damping’s results showed high scatter and no trend was perceived. This is an

area that should be studied in the future; an excitation source with controlled frequency

and intensity is suggested for this purpose.

Vertical load tests

1. All structures had the same behaviour pattern and an elastic response was observed;

2. The values of the elastic moduli estimated with this method were consistent with the

ones obtained in laboratory test specimens, except in the case of the REF structure,

whose shrinkage cracking increased its deflections to such an extent that its equivalent

Young’s modulus (taking into account the loss of stiffness due to stage II cracking) was

lower than the Young’s modulus of B25, when it was expected (after laboratory and dy-

namic response tests) to be higher,

3. An increase of deformations is not an obstacle to the application of these aggregates,

generated from precast elements, for full replacement of the coarse fraction of the natu-

ral aggregates of concrete mixes, since the results showed only a slight decrease in

Young’s modulus (when SP is not used), less significant than factors such as execution

conditions (as proven by the previous conclusion);

4. The use of SP in a total replacement mix proved to be a good option, and the resulting

elastic modulus is the highest of the various mixes tested.

Horizontal destructive load tests

1. All structures behaved with the same pattern, i.e. a ductile collapse mechanism;

2. The load-deformation curves obtained were similar, with no clear impact of recycled

aggregate incorporation;

3. The former conclusions are explained by the relevant role that reinforcement steel has

on the ductile behaviour of structures and on the ultimate and yielding internal forces of

the cross-sections;

4. Execution conditions are more relevant in the structural behaviour than recycled aggre-

gate incorporation, as the similar load-deformation curves and the cracking pattern of

structure B100 suggest;

5. Reinforcement steel is highly effective in reducing any possible detrimental effect of re-

cycled aggregate incorporation in terms of cracking;

6. All structures showed excellent ductility levels.

If code regulations are followed and a proper design is made, recycled aggregate incorporation

is no hindrance to a good seismic behaviour of concrete structures made with them.

20

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the support of Opway, IST, University of Lisbon, the

ICIST research centre and FCT (Foundation for Science and Technology).

REFERENCES

1. Barbudo, A.; de Brito, J.; Evangelista, L.; Bravo, M.; Agrela, F., 2013, “Influence of water-reducing

admixtures on the mechanical performance of recycled concrete”, Journal of Cleaner Production,

Vol. 59, pp. 93-98, Elsevier

2. CEN (European Committee for Standardisation), 2004a, “Eurocode 2 Design of concrete structures.

Part 1-1: General rules and rules for buildings, EN1992-1-1”, Brussels, Belgium

3. CEN (European Committee for Standardisation), 2004b, “Eurocode 7: Geotechnical design. Part 1:

General rules, EN1997-1-1”, Brussels, Belgium

4. CEN (European Committee for Standardisation), 2004c, “Eurocode 8: Design of structures for earth-

quake resistance. Part 1: General rules, seismic actions and rules for buildings, EN1998-1-1”, Brussels,

Belgium

5. Gonçalves, P.; de Brito, J., 2010, “Recycled aggregate concrete (RAC) - Comparative analysis of

existing specifications”, Magazine of Concrete Research, Vol. 62 (5), pp. 339-346, ICE Publishing

6. González-Fonteboa, B., Martínez, F., 2007, “Shear strength of recycled concrete beams”, Construc-

tion & Building Materials, Vol. 21 (4), pp. 887 -893, Elsevier

7. Hansen, T.C., 1992, “Recycling of demolished concrete and masonry. Part one: recycled aggregates

and recycled aggregate concrete”; Report of Technical Committee 37-DRC Demolition and Reuse

of Concrete, pp. 1-139

8. LNEC E 471, 2006, “Guide for the use of coarse recycled aggregates in concrete with hydraulic

binders”, National Laboratory of Civil Engineering, Lisbon, Portugal (in Portuguese)

9. Magalhães, F.; Cunha, A.; Caetano, E.; Brincker, R. (2010), “Damping estimation Using Free Decays

and Ambient Vibration Tests”, Mechanical Systems and Signal Processing, Vol.24, pp. 1274-1290

10. Mander J.B.; Priestley M.J.N.; Park R., 1988,. “Theoretical stress-strain model for confined con-

crete”, Journal of Structural Engineering, Vol. 114 (8), pp. 1804-1826

11. Matias, D.; de Brito, J.; Rosa, A.; Pedro, D., 2013, “Mechanical properties of concrete produced

with recycled coarse aggregates- influence of the use of superplasticizers”, Construction and Build-

ing Materials, Vol. 44, pp. 101-109, Elsevier

12. Pacheco, J.; de Brito, J.; Ferreira, J.; Soares, D., 2014a, “Dynamic characterization of full-scale

structures made with recycled coarse concrete aggregates”, Engineering Structures, submitted for

publication

13. Pacheco, J.; de Brito, J.; Ferreira, J.; Soares, D., 2014b, “Vertical load tests of full-scale recycled

aggregates concrete structures”, Journal of Materials in Civil Engineering, submitted for publication

14. Pacheco, J.; de Brito, J.; Ferreira, J.; Soares, D., 2014c, “Horizontal destructive load tests of full-

scale recycled aggregates concrete structures”, ACI Structural Journal, submitted for publication

15. Pereira, P.; Evangelista, L.; de Brito, J., 2012, “The effect of superplasticisers on the workability and

compressive strength of concrete made with fine recycled concrete aggregates”, Construction and

Building Materials, Vol. 28, pp. 722-729, Elsevier

16. Sato, R.; Mauryama, I.; Sogabe, T.; Sogo, M., 2007, “Flexural behavior of reinforced recycled concrete

beams”, Journal of Advanced Concrete Technology, Vol. 5 (1), pp. 43-61 Japan Concrete Institute

17. Sharma, A.; Reddy, G.R.; Vaze, K.K.; Eligehausen, R., 2013, “Pushover experiment and analysis of a full

scale non-seismically detailed RC” , Engineering Structures, Vol. 46, pp. 218–233, Elsevier

18. Xiao, J.; Li, W.; Fan, Y.; Huang, X., 2012, “An overview of study on recycled aggregate concrete in China

(1996–2011)”, Construction & Building Materials, Vol. 31 (1), pp. 364-373, Elsevier

19. Xiao, J.; Sun, Y.; Falner, H., 2006, “Seismic performance of frame structures with recycled aggre-

gate concrete”, Engineering Structures, Vol. 28 (1), pp. 1–8, Elsevier

20. Wang, C.; Xiao, J., 2013, “Study of the seismic response of a recycled aggregate concrete frame struc-

ture”, Earthquake Engineering and Engineering Vibration, Vol. 12, pp. 669-680, Springer

BEHAVIOUR OF STRUCTURES MADE WITH RECYCLED COARSE ... · behaviour of structures made with recycled coarse aggregates originated from the precast industry joÃo nuno noronha ramos

Documents