·Home ·Table of Contents ·Civil Engineering

Testing Procedure Based on Thermal and Mechanical Analysis for the Evaluation of Concrete Damage Bocca, Pietro Giovanni Crotti, Marco Vincenzo, Di Vasto Contact

Abstract

This paper analyses the procedure of a non-destructive test on existing concrete structures.
The leading principles in the test, supported by concrete specimens drilled in the laboratory, is that the concrete matrix, if it is not damaged, has got a natural cooling behaviour while subjected to compressive loading.
The core drilling method followed in the laboratory very closely simulates real conditions.
The cooling tendency is lower if the specimen had already reached high loading conditions in its practice life. In this case the concrete material is damaged by internal micro-cracks. Harder loading conditions (major than s _test/s _crack=60%) report a higher temperature decrease.
Furthermore the mechanical answer in terms of stress-strain trend (s ,e ) is evaluated in addition to the temperature cooling curves, because even this one appears different between pre-stressed and non pre-stressed specimens.
This methodology is carried out on cylindrical concrete specimens obtained by core drilling on which thermal resistors and electrical displacement transducers are applied.
To end up this procedure allows to offer a simple, valid and forward answer to the concrete damage problem.

Keywords: Concrete, damage, load, cycles, thermal, temperature.

Introduction

This report illustrates an investigation aimed at identifying a method for the evaluation of concrete damage using both mechanical and thermal analyses.
Theoretical considerations were backed up by tests conducted by simulating damage conditions through the application of cyclic loads, including high loads approximating the failure values. This approach made it possible to perform analyses on laboratory made specimens, whose mix was known.
The behaviour of concrete damaged undergoes a twofold variation, as, in addition to mechanical properties, its thermal behaviour is also seen to vary during the application of the compressive loads during the test.
Temperature is seen to decrease if the damage to the specimen being tested has not produced permanent alterations, such as microcracks, in the cement matrix. Otherwise, the temperature of the specimen is seen to increase.
As for the mechanical properties of the material, from an evaluation of the stress-strain curve obtained during cyclic loading tests, we find that the size of the area enclosed by this curve is not constant through the various loading cycles, reflecting changes in the mechanical properties of concrete with increasing damage.
The foregoing concepts can be applied in combined form to carry out tests on existing structures.
The experimental campaign was performed on concrete cores, 60 mm in diameter and 120 mm high, drilled in the laboratory.
The cores were produced by means of a column type drill. The drilling direction was vertical, perpendicular to the direction of casting; the advancement of the drilling machine was obtained by means of continuous manual recall. The method adopted to produce the specimens reproduces quite faithfully the alterations to the material that may be caused by in situ core drilling.

The characteristics of the concrete used in laboratory tests are given in Table 1: they are typical of medium grade concretes as are used in most of the structures that exist and hence may require damage assessments.

	U. of meas.	Quantity
Cem I 42.5 R	Kg/m3	300
Sand 0-4 mm	Kg/m3	883
Fine gravel 5-8 mm	Kg/m3	1040
W/C	-	0.65
Rcm	N/mm2	25
Table 1: Characteristics of the concrete used for the specimens

Thermal behaviour.

In the course of cyclic tests, the temperature of the material tends to decrease in a manner inversely proportional to the working rate already reached and the degree of damage of the specimen.
Figure 1 shows the temperature curves recorded for specimens subjected to loading cycles corresponding to 20% of the failure load (frequency = 1 Hz, s _test/s _fail. = 20%, n = 3600 cycles), with different prior damage conditions.

The curves depicted can be classified as follows:

1. undamaged specimens (not subjected to previous loading cycles);
2. damage specimens (subjected to previous loading cycles, with: frequency = 1 Hz, s _test/s _fail. = 60%, n = 3600 cycles);
3. severely damaged specimens (subjected to previous loading cycles, with: frequency = 1 Hz, s _test/s _fail. = 80%, n = 3600 cycles).

Fig 1: Evolution of temperature in a), b) and c) type specimens.

The tests conducted on specimens displaying different degrees of previous damage made it possible to determine the difference D , in terms of temperature, of undamaged (a) as opposed to damaged specimens (b) and (c).
The higher is the degree of damage reached, the higher is the difference recorded.
The difference (D ₆₀) recorded between a) and b) specimens is markedly lower than the difference (D ₈₀) corresponding to more severe prior damage conditions.
In physical terms, this phenomenon can be accounted for by the opening of microcracks in the cement matrix which produce heat by friction when the specimen is subjected to a compressive load. Undoubtedly, in conditions (c), the phenomenon is enhanced by the fact that damaging process has produced bigger cracks.

Mechanical properties.

The mechanical analysis of loading and unloading cycles during the tests revealed appreciable differences between the first and the last loading cycle; the enclosed area recorded in the initial cycles is large, since part of the energy supplied to the specimen is dissipated in the form of creep (Figure 2).

Fig 2: Stress-strain curves in the first and last cycles of the test.

In the final cycles of the tests, hysteresis is extremely reduced and the behaviour of the material becomes virtually linear-elastic.
If the cyclic tests were continued indefinitely, the areas enclosed by the cycles would start increasing again, on account of the damage to the material (Figure 3).

Fig 3:


It is of interest to note that a specimen's performance characteristics improve as the specimen passes from an initial condition A_p to its end of test condition A_u: the compacting of the material has produced an increase in elastic modulus, i.e., greater stiffness.
Figure 2 shows the stress-strain curves for the first and last cycles of specimens subjected to a cyclic test at 1 Hz, 3600 cycles, s _test/s _failure = 60%.

As can be clearly seen, residual strain is appreciable in the first loading cycle (h _p), decreases as the subsequent loads are applied, approaches zero as the last loads are applied.
A similar behaviour was observed for the area enclosed: large (A_p) for the first cycle, virtually nil (A_u) for the last cycle.
If the test were carried on beyond condition A_u, corresponding to a condition of maximum performance improvement and no major damage, we would go back to cycles with appreciable hysteresis (A_dan) as are typical of "damaged" conditions.

At this point, the process is irreversible and the continuation of the test would inevitably lead to a further weakening of the specimen up to failure.

Application for tests on existing structures.

In the light of the foregoing, i.e., on the basis of the investigations performed and the literature available on these issues [1,2], it proves possible to define a testing method for use on the concrete of existing structures.
The steps of the testing method can be summarised as follows:
1. producing a representative sample consisting of cores with a diameter proportional to maximum aggregate diameter (f _core > 2 f _{max, aggregate}).

By way of exemplification, a core-drilling scheme in embankment wall facing is described in Figure 4. The samples taken from zone A) were damaged solely by atmospheric agents. Those from zone B) were also damaged by compressive loads, including some of a cyclic nature under the assumption of accidental overloads being applied upstream of the wall (e.g. a road).

Fig 4: Example of drilling in a embankment wall

2. determining the mechanical characteristics of the specimens, such as elastic stiffness E*, failure stress s _r and behaviour at the post-failure stage (Figure 5).

Fig 5: Mechanical properties recorded during the test

3. performing a cyclic loading test at 20% of s _r, frequency = 1 Hz, n = 3600 cycles, and at the same time measuring the temperature of the material by means of resistors applied to the specimen (Figure 6).

Fig 6: Evolution of temperature during cyclic tests

4. performing a cyclic loading test at a maximum pre-determined load value of X, with the following testing parameters: s _test/s _failure = X%, frequency = 1 Hz, n = 3600 cycles.

By comparing the first and the last cycle of the test we may distinguish two conditions:

4.1) h _u @ h _p; A_u @ A_p

We may infer that during its service life the specimen had already reached a degree of damage corresponding to the cyclic test with X% loading. At this point, it proves necessary to compare the results with those obtained from the 20% cyclic test conducted by monitoring the evolution of temperature. If the temperature has increased greatly (D value) compared to that of the reference specimens, this means that during its service life the specimen had undergone a degree of damage corresponding to, or greater than, the X load: the cement matrix was altered accordingly and microcracks opened resulting in a proportionally appreciable development of heat.
The test is reiterated by increasing the load up to (X+i)%. Only when the (X+i)% load is reached and the residual return strain of the last cycle, h _u, is seen to be considerably lower than the residual strain of the first cycle, h _p, the same condition applying to A_u and A_p, it can be concluded that the specimen had never reached the corresponding damage before.

4.2) h _u << h _p

The specimen in question has never reached a damage corresponding to the X% load. The test is reiterated by reducing the load to (X-i)%. As soon as the load (X-i)% is seen to give rise to a residual return strain of the last cycle, h _u, corresponding to the residual return strain of the first cycle, h _p, the same condition applying to A_u and A_p, we may conclude that the specimen had already reached the corresponding damage before.


The comparison with the evolution of temperature in 20% cyclic tests is necessary not so much in order to evaluate the load reached in absolute terms, as to assess the egree of damage reached by specimen by determining to what extent the curve of the specimens tested deviates (D ₆₀, D ₈₀) from the reference condition (20% s _r).

Conclusions.

The proposed method, making use of both thermal and mechanical analyses, instead of mechanical analysis alone, supplies additional cognitive parameters which, in many instances, prove decisive in the determination of damage conditions. Compared to other tests involving costlier and more awkward procedures for the production of test pieces, this method supplies a more accurate and wider spectrum reading of the actual conditions of the concrete being studied. The basic principle underlying the thermal tests is that damaged concrete contains microcracks resulting in the development of heat when the material is subjected to cyclic loading.

Acknowledgements

The contribution of MURST is acknowledged for the project on "Structural integrity assessment of large dams".

References.

1. Ballatore E., Bocca P., Variations in the mechanical properties of concrete subjected to low cyclic loads, in Cement and concrete research, vol. 27, pp. 453-462, USA, 1997
2. Berra M., Bocca P., Thermoelastic stress analysis: temperature-strain relationship in concrete and mortar, in Materials and structures, V. 26, pp. 395-404, Paris, 1993
3. Constantin Avram, Concrete strength and strains, Elsevier, Romania, 1981
4. Collepardi M., Scienza e tecnologia del calcestruzzo, Hoepli, Milan, 1980

© AIPnD , created by NDT.net

|Home|    |Top|