Professional Documents
Culture Documents
A Joint Fatigue-Creep Deterioration Model For Masonry With Acoustic Emission Based Damage Assessment
A Joint Fatigue-Creep Deterioration Model For Masonry With Acoustic Emission Based Damage Assessment
A Joint Fatigue-Creep Deterioration Model For Masonry With Acoustic Emission Based Damage Assessment
h i g h l i g h t s
Development of SN and ST type failure prediction models for fatigue and creep in masonry, based on experimental results.
Characterisation of stages of the damage accumulation process with acoustic emission techniques.
A joint deterioration model for fatigue and creep in masonry.
a r t i c l e i n f o a b s t r a c t
Article history: The paper investigates the long-term fatigue and creep deterioration processes in historical brick
Received 29 October 2012 masonry. Based on two independent laboratory test series, the relationship between stress level and life
Received in revised form 14 February 2013 expectancy was considered for fatigue and creep loading in the form of SN type models. The process of
Accepted 26 February 2013
deterioration was investigated with the help of acoustic emission technique to identify stages and char-
Available online 3 April 2013
acteristics of the damage accumulation process. Based on the test data and acoustic emission results, a
joint SN type deterioration model was proposed to incorporate the static, fatigue and creep deterioration
Keywords:
mechanisms. A mathematical relationship was proposed for the joint fatigue–creep model and good
Fatigue
Creep
agreement was found between the test data and the proposed model.
Masonry Ó 2013 Elsevier Ltd. All rights reserved.
Acoustic emission technique
Failure prediction
Joint deterioration model
0950-0618/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.conbuildmat.2013.02.045
576 A. Tomor, E. Verstrynge / Construction and Building Materials 43 (2013) 575–588
2. Experimental approach
Brick masonry prisms (Fig. 3) have been built and tested at the University of the
West of England (UWE, Bristol, UK) under long-term fatigue loading to help develop
understanding of the performance of masonry arch bridges under long-term traffic
loading. The prisms were built using solid 213 100 65 mm3 Wienerberger
Warnham Red Terracotta moulded bricks with 22.6 N/mm2 strength and 2127 kg/
m3 density. 1:1:6 cement:lime:sand mortar was used by volume with NHL3.5 lime
and 3 mm sharp washed sand. Joint thickness was 8 mm and specimens were cured
for a minimum of 6 month before testing. The average compressive strength of
prisms (SAv) was 10.9 N/mm2 (1.0 N/mm2 standard deviation (SD) and 9.3% coeffi-
cient of variation (CV)), see Table 1.
Prisms were tested under compression using a 250 kN actuator at 0.15 kN/s
loading rate. Layers of 3 mm plywood and 30 mm thick steel plates were placed
on top and bottom of the specimens for load distribution.
10 prisms were tested under static loading to identify the average compressive
strength (SAv). Long-term cyclic loading was subsequently applied to further spec-
imens at 2 Hz frequency between a minimum stress level (SMin) and a maximum
stress level (SMax), that were defined as percentage of the average static strength
(SAv), see Fig. 4.
The minimum stress is intended to represent the dead load of a bridge due to its
self-weight and the maximum stress the variable live loads induced by passing traf-
Fig. 2. Creep damage at the base of the bell tower of Saint-Willibrordus church, fic. The minimum stress (or base load) was applied either at 10% or 30% of the aver-
Belgium, a few weeks before collapse. age static strength (SAv) during the tests.
A. Tomor, E. Verstrynge / Construction and Building Materials 43 (2013) 575–588 577
Fig. 3. Prisms with instrumentation (a), before (b) and after (c) testing.
Table 1
Overview of material characteristics of bricks and masonry for fatigue tests. Values are indicated with average strength (SAv), standard deviation (SD) and coefficient of variation
(CV).
(a) for Short-Term Accelerated Creep Tests (short-term ACT) the duration of the
2.2. Setup for creep testing loading steps (Dt) was 3 h and failure generally occurred within 24–36 h.
(b) for Long-Term Accelerated Creep Tests (long-term ACT) the duration of the
A similar setup was used in the Reyntjens Laboratory at the Catholic University loading step (Dt) was 2 months and failure generally occurred within 1.5–
of Leuven (KU Leuven, Belgium) to test the creep behaviour of masonry under uni- 2 years for each specimen.
axial compressive loading. (c) Additionally, 1-Step Creep Tests (1-SCT) were performed to provide data for
Relatively low-strength solid 188 88 48 mm3 Terca ‘Spanish Red’ clay damage accumulation at specific stress levels. During the 1-step tests the
bricks and 1:2.5 lime:sand mortar was used by volume with hydrated lime and load was increased up to 80–90% of the average compressive strength
2 mm sand. Both mortar type and composition were chosen to be representative within half an hour and kept constant until failure occurred. By maintaining
of historical masonry in Western Europe. Strength characteristics of the masonry the stress level for an extended period of time, the tertiary creep phase had
components and full scale specimens are listed in Table 2. more time to develop and the likelihood of failure during stress increase
Masonry columns with overall dimensions 188 188 600 mm3 was reduced. A constant stress level of 80–90% of the average strength
(length width height) were built with 10 layers of bricks (two bricks per layer) (SAv) was chosen based on experience, to allow the tertiary creep phase
and 10 mm mortar joints as shown in Fig. 5. Each specimen was constructed on a to develop within 3–24 h.
concrete tile and a similar concrete tile or a steel plate was placed on top of the col-
umns after 1.5 months. Mortar samples and masonry specimens were stored at a
Quasi-static compressive tests and short-term creep tests were performed in a
temperature of 20 °C and relative humidity of 60 ± 5% for three months before test-
Dartec hydraulic press (Fig. 5b) with 5000 kN capacity and an additional loadcell to
ing. The lime mortar was fully carbonated by storing the mortar and masonry spec-
limit the maximum load to 500 kN and increase accuracy. Long-term and 1-step
imens in a carbonation chamber with high CO2 level for several weeks.
Table 2
Overview of material characteristics of bricks, mortar and masonry for creep tests. Values are indicated with average strength (SAv), standard deviation (SD) and coefficient of
variation (CV).
Fig. 5. Test specimen, (a) short-term (b) long-term and (c) creep test setup.
During the fatigue test series at the University of the West of England, acoustic
emission monitoring was performed using a Physical Acoustics Micro-SAMOS sys-
tem with eight-channel PCI-8 AE board connected to PC via PCMCIA port and
AEwin™ software was used to process the data. IL40s voltage preamplifiers with
1–400 kHz frequency bandwidth and 40 dB gain were used. 150 kHz resonant
R15 sensors with 50–200 kHz operation frequency were attached to opposite sides
of the specimens by means of a thin layer of hot-melt glue, which has proven to be a
good couplant for laboratory conditions. The AE system was calibrated using the
standard method of pencil lead breaks [44] for every test to verify the sensitivity
of the sensors. Threshold level was adjusted during each test to avoid saturation.
Fig. 6. Loading pattern for stepwise short-term and long-term accelerated creep
AE monitoring was performed during each of the static tests and during some of
tests (ST-/LT-ACT) and for 1-Step Creep Tests (1-SCT).
the fatigue tests that failed within a relatively short time interval, as long-term tests
exceeded the capacity of the data logger.
During the creep test series at the KU Leuven, acoustic emission monitoring was
creep tests were performed in individual steel frames for each specimen (Fig. 5c)
performed using a 4-channel Vallen AMSY-5 system. 375 kHz resonance sensors
due to the extended time requirements (up to 2 years). The latter setup closely
with 250–700 kHz operation frequency were attached to opposite sides of the spec-
resembled the loading and boundary conditions of the short-term test setup. The
imens by means of a thin metal plate that was glued to the surface and allowed easy
load was applied and increased manually using a hydraulic jack with an accumula-
re-mounting for periodic monitoring. Vacuum grease was used as a couplant be-
tor to compensate for relaxation and keep the loading constant. All creep tests were
tween the sensor and the metal plate. The preamplifier gain was set to 34 dB with
performed under load-control.
a fixed threshold level of 34.5 dB. High frequency noise was filtered by applying a
low-pass filter at 500 kHz. Pencil lead breaks were used for system calibration
2.3. Instrumentation (strain and acoustic emission monitoring) [44] and could be detected up to 30 cm from the sensor, indicating that almost
all damage sources could be detected by only two sensors in the middle of the spec-
During both test series Linear Variable Differential Transformers (LVDTs) were imen. During the quasi-static compression tests, short-term creep tests and 1-step
used to monitor deformation. During fatigue tests deformation was measured creep tests, AE events were continuously monitored. For long-term creep tests, AE
across two mortar joints. Under short-term creep tests axial and lateral deformation sensors were attached 1 day before stress increase and remained in position for
was constantly recorded on all four sides of the specimens using LVDTs and under 6 days after stress increase. If unstable or increasing damage accumulation was ob-
long-term creep tests deformation was recorded periodically using a Demountable served after 6 days, the monitoring period was extended. Additional AE monitoring
Mechanical DEMEC strain gauge. was performed for 24 h halfway between two loading steps to identify if emission
In addition to deformation monitoring, the Acoustic Emission (AE) technique rate was stable.
was used during both test series to help identify damage development characteris-
tics under fatigue and creep loading. The technique has the advantage that it re- 2.4. Experimental setup: similarities and differences
cords the structure’s response to loading in real time. Acoustic emission
technique detects high-frequency transient elastic waves that are emitted by the It is evident that the two test series for fatigue and creep deterioration were not
material itself during crack growth [41,42]. Waves are recorded on the surface by part of the same research program and differed in the choice of materials and test-
piezoelectric sensors, pre-amplified, filtered and amplified before they are pro- ing arrangements. However, despite of the differences, they had common goals and
cessed by the data logger. AE amplitude is detected in lV, converted to AE decibel methodologies. Both fatigue and creep are time-dependent deterioration processes
by 1 dB = 20 log (Voltage (lV)/1 lV) [43] and energy is calculated as the area un- at stress levels below the maximum strength of the material. While fatigue is in-
der the envelope of an AE hit. Background noise is eliminated through a minimum duced by variable stress levels, creep may be considered as fatigue loading with
amplitude threshold. For each AE hit a number of parameters (e.g. amplitude, en- zero stress amplitude. It is therefore expected that common features may be ob-
ergy, duration, count, arrival time) and the waveform are recorded. If an AE hit is served between the two phenomena that can help develop greater understanding
recorded by one or more sensors it is defined as an AE event. of the damage evolution process in masonry.
The amount of detected AE hits and energy is influenced by the applied hard- Before discussing the test results in detail, differences between the two test ser-
ware and software (e.g. sensor type and frequency range, applied filters, waveform ies are summarised below.
sampling frequency, threshold level, hit duration discrimination time, etc.). There-
fore, software defined parameters should be kept constant for subsequent tests and The masonry composition, specimen geometry (stacked bricks and bonded
results obtained with different acquisition systems should be compared carefully bricks) and compressive strength of the fatigue and creep specimens were dif-
and in relative terms. The amount of AE hits detected during a time interval is also ferent (10.9 N/mm2 and 3.73 N/mm2 respectively). Therefore, relative stress (in
sensitive to a number of setup-specific boundary conditions, such as quality of the % of SAv) rather than absolute stress will be used for comparison. Damage evo-
coupling between sensor and test specimen, material density, speed of wave prop- lution during creep has been found to follow similar patterns for similar
agation, interference with surrounding test equipment, and presence of internal masonry types if compared in terms of relative stresses [2].
cracks and voids. Due to the variability of the boundary conditions, relative change Strain was measured differently: axial strain across 2 bed joints during fatigue
in detection level is therefore better suited for defining damage accumulation in- loading and lateral and axial strain over 1 and 3 or 5 bed joints during creep
stead of the absolute amount of hits/energy. loading. Strain results will therefore not be directly compared in this paper.
A. Tomor, E. Verstrynge / Construction and Building Materials 43 (2013) 575–588 579
Different AE systems and sensors were used that are likely to influence the mag- With the help of AE technique, the process of micro crack for-
nitude and properties of detected acoustic emission events. AE in masonry is mation was detected from the very onset during compressive tests
however also affected by a series of further factors, such as system setup, inho-
mogeneities of the material, relative diffraction, reflection and attenuation of
on masonry prisms. Typical acoustic emission amplitude and aver-
brick and mortar, etc. As direct quantitative comparison of the two test pro- age energy recordings, obtained under quasi-static compressive
grammes would be unreliable, qualitative analysis will be used to identify loading of prisms, are shown in Fig. 7a and b. Noticeable emission
changes in emission patterns and indicate damage evolution. was recorded from almost the start of load application and clear in-
crease was observed around 40% and 95% of the maximum stress
(SMax) in both amplitude and average energy.
3. Deterioration process
The amplitude threshold was set to 40 dB at the beginning of
the test and was progressively raised to avoid saturation (loss of
Deformation in masonry is a consequence of elastic and plastic
linearity between input and output in the amplifier due to exces-
strain, micro-cracking and pore collapse within the material (the
sive signal drive). Although low-amplitude hits were not recorded
latter phenomena producing acoustic emissions). Before deforma-
under high loads, they were considered to be less significant in the
tion can be measured on the surface of the masonry, AE technique
presence of high intensity signals and have negligible effect on the
can generally capture emissions due to micro-crack development
energy output (shown on the log scale).
and crack growth. Based on the AE recordings under quasi-static
Based on the AE recordings, three phases of the crack develop-
compressive loading, fatigue loading and creep loading, character-
ment can be identified during quasi-static compression (marked as
istics of the deterioration processes were investigated to identify
‘‘S-Phase’’ for ‘‘Static-Phase’’), such as:
phases of damage development. Thereby, a parameter-based tech-
nique is adopted, analysing AE parameters such as amplitude, en-
S-Phase 1: relatively constant low-level emission is observed,
ergy and event-counting to assess the damage accumulation
likely to be associated with compaction and crack nucleation,
within the masonry specimens. Other techniques, involving a more
while elastic strain is the dominant cause of deformation at this
complex signal-based analysis were developed for concrete [41],
stage.
but have not yet been applied on masonry due to the very complex
S-Phase 2: micro-crack development from around 40% of SMax,
fracture behaviour of masonry under compression and the hetero-
likely to be associated with development of vertical cracks,
geneity of the material.
crack extension and further compaction of the mortar. Defor-
mation of the mortar induces local stress concentration in the
adjacent bricks with increasing plastic behaviour and develop-
3.1. Deterioration under static loading
ment of micro-cracks in the bricks. S-Phase 2 appears to have
a steeper initial section (a) with rapid amplitude and energy
It is generally accepted that macroscopic fracture in brittle
increase, followed by somewhat reduced amplitude and energy
materials under uniaxial compression is caused by nucleation,
release (b), but the exact mechanisms in the two sub-phases are
growth, interaction and coalescence of micro cracks [45,46]. Micro
yet to be identified
fracture originates at flaws in the material, such as grain bound-
S-Phase 3: very high energy release from around 95% of SMax,
aries, and the stress at the crack tip responsible for crack growth
characterised by macro-cracking and bridging while the tensile
is of tensile nature. Consequently, cracks propagate in the direction
strength of bricks is being reached and fast fracture occurs in
parallel to the principal compressive stress. In masonry, the inter-
the bricks.
action between bricks and mortar, which have different stiffness
properties, is assumed to contribute to the concentration of stres-
In order to gain better insight into the fracture development
ses and subsequent crack formation. In the general case where the
process of masonry components, half size bricks (105 100
brick units have a higher stiffness than the mortar, deformation of
65 mm3) and 100 100 100 mm3 (1:1:6) mortar cubes were
the mortar induces local tensile stress concentration in the adja-
subsequently tested under quasi-static compression (Fig. 8a and
cent bricks, while enclosed mortar joints are subjected to triaxial
b). For bricks, the AE amplitude and average energy increased rap-
compressive stresses [47].
Fig. 7. Typical AE amplitude (a) and average energy (b) vs. stress during compression of a masonry prism.
580 A. Tomor, E. Verstrynge / Construction and Building Materials 43 (2013) 575–588
Fig. 8. Typical AE amplitude and average energy during compression of a brick (a and b) and mortar cube (c and d).
idly between 0% and 30% SMax (up to ca. 90–100 dB) and a very high 3.2. Deterioration under fatigue loading
emission level was maintained as the bricks disintegrated between
30% and 100% SMax. Elastic deformation and micro-cracking are 3.2.1. Test results under fatigue
likely to have taken place very early on during the loading process After the static loading tests, prisms were subjected to long-
and macro-cracking is likely to occur from as early as 30% SMax due term cyclic loading to investigate the fatigue deterioration process.
to the brittle nature of the bricks. For mortar, AE amplitude and en- Sinusoidal cyclic loading was applied at 2 Hz frequency for a min-
ergy remained relatively low (40–60 dB) up to ca. 95% SMax, fol- imum of 3,000,000 load cycles unless failure occurred. Loading was
lowed by sudden increase and immediate failure (Fig. 8c and d). however continued to several million cycles in most cases and was
During compressive loading of prisms, changes in energy levels only stopped due to excessive time requirements.
may therefore be associated with the following mechanisms: A typical example of acoustic emission recording during a fati-
gue test is shown in Fig. 9 for a loading range between 10% (=SMin)
S-Phase 1: low-level emission (40–60 dB and ca. 103 aJ, Fig. 7), and 70% stress (=SMax). During the fatigue deterioration process
associated with compression of the mortar joints. three fatigue phases (‘‘F-Phase’’ for ‘‘Fatigue-Phase’’) have been
S-Phase 2: medium-level (60–85 dB and 103–106 aJ), crack distinguished:
extension in the mortar joints is likely to be overridden by the
micro-crack development in the bricks. F-Phase 1: (0–75% of the total number of cycles), relatively low,
S-Phase 3: high-level emission (80–90 dB and 106–107 aJ), likely constant emission (40–50 dB amplitude and 10–102 aJ absolute
to be almost entirely associated with macro-cracking of bricks. energy);
F-Phase 2: (75–95% cycles), small increase in emission (50–
When comparing the AE output for prisms, bricks and mortar 60 dB amplitude and 10–102 aJ absolute energy);
cubes in terms of amplitude level, the micro-cracking and critical F-Phase 3: (95–100% cycles), rapid increase in emission and
crack propagation seem to have initiated when ca. 60 dB was ex- sudden failure. Brief warning period.
ceeded. 60 dB therefore appears to be an approximate critical limit
for damage development for the applied test setup in all three While it is difficult to make specific assumptions about the fati-
cases presented. gue deterioration process of prisms at this stage, it may be helpful
A. Tomor, E. Verstrynge / Construction and Building Materials 43 (2013) 575–588 581
Fig. 9. Typical AE amplitude (a) and average energy (b) vs. number of cycles during fatigue compression of a masonry prism.
Fig. 13. Typical long-term accelerated creep test with stepwise load application (DT = 2 months): AE event rate vs. time for 41% (a), 82% (b) and 95% of SAv (c).
If creep failure occurs within the constant stress interval, ter- creep tests. For these latter tests, the load history is not taken into
tiary creep can clearly be distinguished in the AE recordings account and only the stress level and associated time between
(see Fig. 11a for the 1-step and Fig. 13c for the long-term creep stress increase and failure in the last loading step is shown. The re-
tests). sults therefore do not take into account damage accumulation in
preceding loading steps, thus possibly underestimating the life
Figs. 11 and 12 presented AE amplitude levels for 1-step and expectancy.
short-term accelerated creep tests. During stress increase and in
C-Phase 1, relatively high AE amplitudes can be observed. The low- 3.3.3. Mathematical model for creep
er AE amplitudes (50–60 dB) during C-Phase 2 can be related to For creep loading, an analytical relationship between stress le-
compaction of the mortar and possibly a very slow micro-cracking vel and time to failure has been suggested by Verstrynge [2] as
process in the mortar and bricks, comparable with fatigue F-Phase shown in Eq. (4).
2. In C-Phase 3, an increase in AE amplitude accompanies macro-
crack development within the bricks. As the specimen composition ð1 ðA S þ BÞÞnþ1
T¼ ð4Þ
and AE acquisition systems differ for the creep and fatigue tests cðn þ 1Þ Sn
series, a quantitative comparison of the AE amplitude and energy
emission levels will however not be made. where T is time to failure and S is the ratio of the maximum
stress to the average strength (S = SMax/SAv). Parameters A = 1.9,
B = 0.9, c = 8.5 1011 and n = 8 were calibrated based on the
3.3.2. ST curves under creep creep test results reported in this paper, for which the calibration
Similarly to the SN curves for fatigue loading, the relative stress process can be found in [2]. The model is presented in Fig. 14 to-
level (S in % of SAv) and time to failure (T) may be expressed as ST gether with the experimental creep test results and good corre-
curves for creep loading. Fig. 14 shows the creep test results for the spondence is observed.
20 specimens for which a tertiary creep phase has been observed.
Time to failure (T) is defined as the time between stress increase 3.3.4. AE–T curves under creep
and creep failure. There is a large scatter in the test results simi- While Fig. 14 shows damage development in terms of stress le-
larly to the fatigue tests that may be explained by similar reasons. vel and time to failure, crack development may also be expressed
A logarithmic trendline is indicated with upper and lower limits in terms of AE activity against time. As higher stress levels are
defined by the coefficient of variation (12.6%) of the static com- likely to induce higher intensity micro- and macro-crack develop-
pression tests. ment, strong correlation is expected to exist between stress level
It needs to be noted, that creep test results include not only the and AE intensity [37]. In this respect, the 3-phase creep curve
1-step creep tests, but also the short- and long-term accelerated and more specifically the relationship between the rate of deterio-
ration during the secondary creep phase and stress level can be de-
scribed with a Weibull distribution model. This statistical
background for creep failure prediction based on AE detection
was presented by Verstrynge et al. in [37]. Fig. 15 shows the AE
event rate (number of AE events per minute) during the secondary
creep phase against time to failure (T) for the same set of test re-
sults as Fig. 14. An improved relationship can be observed in the
AE–T graph compared to the ST graph and can help identify the
time to failure based on monitored AE event rate. In Fig. 15, values
are presented on a double logarithmic scale that implies decreasing
accuracy for increasing time to failure. Also, prediction accuracy
may reduce for lower stress levels (longer times to failure) if the
AE activity is of similar magnitude as the background noise.
Although the values in the AE–T curve are only indicative of a spe-
cific masonry type and AE setup, tested under constant environ-
mental conditions, it intends to demonstrate the general
Fig. 14. ST curve for compressive creep tests: stress (S) vs. time to failure (T, log principle for prediction of creep failure in masonry based on AE
scale) for specimens for which a tertiary creep phase was observed. detection.
A. Tomor, E. Verstrynge / Construction and Building Materials 43 (2013) 575–588 585
Fig. 16. Joint failure model indicating stress vs. time or cycles to failure (ST or SN curves) for static, fatigue and creep loading.
586 A. Tomor, E. Verstrynge / Construction and Building Materials 43 (2013) 575–588
4.2. Joint fatigue–creep mathematical model test program are presented as creep tests at 100% mean stress with
0% amplitude.
For relating the experimental test results to mathematical mod- In an environment where ambient fluctuations cannot be con-
els, the curves indicating fatigue failure for a given base load (SMin) trolled, fatigue and creep deterioration are likely to interact with
are simulated by means of the fatigue model adapted from Casas other physical, physico-chemical and biological deterioration pro-
[49] as described in Section 3.2.3. The creep curve (upper SAmpl cesses, making the prediction of life expectancy purely based on
curve) is simulated by means of the creep model described in stress level increasingly unreliable. To the authors’ knowledge no
Section 3.3.3. extensive experimental test programs have been reported in the
While the fatigue model by Casas shows reasonable agreement literature to date on the interaction between fatigue, creep and
with the fatigue test data at 10% minimum stress (Fig. 10), no ade- other deterioration phenomena. The influence of moisture ingress
quate correlation has been found between the fatigue curves and on creep failure of sandstone and the effects of moisture move-
the creep model for higher SMin values. For an amplitude value of ment and salt crystallization on creep strain have however been
zero, SMin, SMean and SMax become equal for a specific fatigue curve studied [3,24].
and the number of cycles to failure should correspond to the failure Although the available test data is insufficient at this stage to
time indicated by the creep model at that specific stress level. In fully quantify and validate the SN relationship, they are intended
order to improve the relationship, the fatigue model by Casas Eq. to be used to demonstrate the methodology and partially validate
(2) has been adapted and a correction factor (C) introduced Eq. (5). the proposed fatigue-creep model for masonry.
S ¼ A NBð1CRÞ ð5Þ
5. Conclusions
where S is the ratio of the maximum stress to the average
strength (S = SMax/SAv), N the number of cycles, R the ratio of the Fatigue and creep deterioration are time-dependent mecha-
minimum stress to the maximum stress (R = SMin/SMax), parameter nisms that can lead to failure over time even at relatively low
A is set to 1, parameter B is set to 0.04 and C is the correction factor. stress levels. Fatigue and creep loading were considered for ma-
To achieve the best correlation with the current fatigue and creep sonry in terms of deterioration process, life expectancy, SN or ST
test results, the value of 0.62 has been identified for parameter C. curves and mathematical models in order to develop a joint fati-
The correction factor allows the interaction between the creep gue–creep deterioration model.
and fatigue phenomena to be taken into account and the slope of Fatigue and creep deterioration were tested within two inde-
the SN curves to be adjusted. pendent laboratory test series at two universities with slightly dif-
This correction for higher stress levels might be explained by ferent test setups, however results were comparable in terms of
the interaction between creep and fatigue, as the influence of creep relative stress and qualitative changes in the acoustic emission
damage is expected to become more significant for higher stress output.
levels. Consequently, the slope of the fatigue curve and the number The acoustic emission technique has shown to be a valuable
of cycles to failure will be slightly reduced due to creep effects. In tool for damage detection during creep and fatigue testing. With
terms of fatigue and creep effect, creep deterioration is assumed to the help of acoustic emission monitoring, stages of the deteriora-
be the dominant cause of failure for higher base loads, while fati- tion process (micro-crack nucleation, growth, and coalescence into
gue damage becomes more dominant for lower mean stresses. macro-cracks) were identified and characterised during static, fati-
The experimental fatigue and creep test data, modified fatigue gue and creep loading.
Eq. (5) and creep model Eq. (4) are jointly shown in Fig. 17. Good Close relationship between AE signal characteristics (amplitude,
correlation is observed between the test data and proposed modi- energy and event count) and remaining life expectancy has been
fied model. To relate the results of the fatigue and creep tests with found during laboratory testing that can help predict the remain-
their respective static tests, the static tests of the fatigue testing ing service life during laboratory and field testing.
with 10% base load are included as fatigue tests with 1 cycle, 55% Based on the fatigue test results, SN curves (stress vs. number of
mean stress and 45% amplitude, while the static tests of the creep cycles) have been developed for a specific masonry type and indi-
90% SMin
80% SMin
70% SMin
60% SMin
0%SAmpl
50% SMin 0%Δ S
40% SMin
5%SAmpl
30% SMin 10%Δ S
45%SAmpl
20% SMin
90%Δ S 40%SAmpl 10%SAmpl
80%Δ S 35%SAmpl 10% SMin 20%Δ S
70%Δ S 30%SAmpl 0% SMin
60%Δ S 25%SAmpl 15%SAmpl
50%Δ S 20%SAmpl 30%Δ S
40%Δ S
Fig. 17. Joint failure model indicating stress vs. cycles to failure for static, fatigue and creep loading, with experimental data.
A. Tomor, E. Verstrynge / Construction and Building Materials 43 (2013) 575–588 587
cated very significant changes in life expectancy for relatively [17] Brencich A, de Felice G. Brickwork under eccentric compression: experimental
results and macroscopic models. Constr Build Mater 2009;23(5):1935–46.
small changes in stress levels. A fatigue model was proposed and
[18] Ronca P, Franchi A, Crespi P. Structural failure of historic buildings: masonry
showed good correspondence with experimental data. fatigue tests for an interpretation model. In: Modena C, Lourenço PB, Roca P,
Creep test results were summarised in an ST relationship (stress editors. 4th International conference on structural analysis of historical
vs. time) and an existing creep model with good correlation with constructions. Taylor and Francis: Padova, Italy; 2005. p. 273–279.
[19] Tomor AK, Wang J. Fracture development process for masonry under static and
the experimental data was applied. An AE-T relationship (AE event fatigue loading. In: 6th International conference on arch bridges: Fuzhou.
rate vs. time to failure) was found to provide improved accuracy China; 2010. p. 550–560.
for estimating the time to failure and rate of damage propagation. [20] Wang J, Tomor AK, Melbourne C, Yousif S. Critical review of research on high-
cycle fatigue behaviour of brick masonry. Constr Build Mater, in press.
A joint SN model was proposed to incorporate static, fatigue and [21] Melbourne C, Tomor AK, Wang J. Cyclic load capacity and endurance limit of
creep loading in terms of stress against time or number of cycles to multi-ring masonry arches. In: 4th Arch bridges conference: Barcelona; 2004.
failure. Creep was considered as fatigue loading with zero ampli- p. 375–384.
[22] Melbourne C, Tomor AK, Wang J. Modes of failure of multi-ring masonry
tude and static loading as fatigue failure after one cycle. arches under fatigue loading. In: 5th International conference on bridge
The proposed fatigue model was adapted to improve correlation management: Surrey. UK; 2005. p. 476–483.
with the creep model and both mathematical models were com- [23] Melbourne C, Wang J, Tomor AK. A New Masonry Arch Bridge Assessment
Method (SMART). In: ICE – Bridge Engineering; 2007. p. 81–87.
bined in a fatigue–creep deterioration graph. Good agreement [24] Forth JP, Brooks JJ. Creep of clay masonry exhibiting cryptoflorescence. Mater
was found between the test data and proposed model for the Struct 2008;41(5):909–20.
tested time span. No endurance limit was observed within the [25] Forth JP, Brooks JJ, Tapsir SH. The effect of unit water absorption on long-term
movements of masonry. Cement Concr Compos 2000;22(4):273–80.
timeframe of the experimental tests.
[26] Lenczner D. Creep and prestress losses in brick masonry. Struct Eng
Further laboratory tests data are required to populate the pro- 1986;64B(3):57–62.
posed fatigue–creep model and allow a generic quantitative model [27] Anzani A, Binda L, Mirabella Roberti G. The effect of heavy persistent actions
to be developed for practical use. Also, further work is required to into the behaviour of ancient masonry. Mater Struct 2000;33(228):251–61.
[28] Ignoul S, Schueremans L, Tack J, Swinnen L, Feytons S, Binda L, et al. Creep
quantify the effects of external influences and fluctuating environ- behavior of masonry structures – failure prediction based on a rheological
mental conditions on creep and fatigue deterioration and on AE model and laboratory tests. In: Lourenço PB, et al., editors, 5th International
emission levels to enable practical life-cycle prediction. seminar on structural analysis of historical constructions. New Delhi; 2006. p.
913–920.
[29] Verstrynge E, Schueremans L, Van Gemert D. Time-dependent mechanical
behavior of lime-mortar masonry. Mater Struct 2010;44(1):29–42.
Acknowledgements [30] Melbourne C, Tomor AK. Application of acoustic emission for masonry arch
bridges. Strain 2006;42(3):165–72.
[31] De Santis S, Tomor AK. Laboratory and field studies on the use of acoustic
The authors gratefully acknowledge the financial support pro-
emission for masonry arch bridges (<http://dx.doi.org/10.1016/
vided by the Engineering and Physical Research Sciences Council j.ndteint.2013.01.00.6i>). Ndt & E International 2013.
(EPSRC) for the research project at UWE and by the Research Foun- [32] Masera D, Bocca P, Grazzini A. Frequency analysis of acoustic emission
dation – Flanders (FWO) for the postdoctoral research of E. signal to monitor damage evolution in masonry structures. In: 9th
International conference on damage assessment of structures (Damas
Verstrynge. 2011); 2011. p. 305.
[33] Tomor AK, Melbourne C. Condition monitoring of masonry arch bridges using
acoustic emission techniques. Struct Eng Int 2007;2:188–92.
References [34] Shigeishi M, Colombo S, Broughton KJ, Rutledge H, Batchelor AJ, Forde MC.
Acoustic emission to assess and monitor the integrity of bridges. Constr Build
Mater 2001;15(1):35–49.
[1] Binda L, editor. Learning from failure – long-term behaviour of heavy masonry
[35] Invernizzi S, Lacidogna G, Manuello A, Carpinteri A. AE monitoring and
structures. Advances in architecture, vol. 23. Southampton: WIT Press; 2008.
numerical simulation of a two-span model masonry arch bridge subjected to
[2] Verstrynge E. Long-term behaviour of monumental masonry constructions:
pier scour. Strain 2011;47:158–69.
modelling and probabilistic evaluation. PhD thesis. civil engineering
[36] Carpinteri A, Invernizzi S, Lacidogna G. AE structural assessment of a 17th
department, K.U.Leuven: Leuven; 2010.
century masonry vault. In: Lourenco PB, et al., editors. 5th International
[3] Verstrynge E, Konings S, Wevers M. The influence of moisture on creep
seminar on structural analysis of historical constructions. New Delhi; 2006.
behaviour of sandstone assessed by means of acoustic emission. In: Jasienko J,
[37] Verstrynge E, Schueremans L, Van Gemert D, Wevers M. Monitoring and
editor. 8th International conference on structural analysis of historical
predicting masonry’s creep failure with the acoustic emission technique. NdtE
constructions. Wroclaw, Poland; 2012. p. 2573–2581.
Int 2009;42(6):518–23.
[4] Ferretti D, Bazant ZP. Stability of ancient masonry towers: moisture diffusion,
[38] Verstrynge E, Schueremans L, Van Gemert D. Predicting the time to failure in
carbonation and size effect. Cement Conc Res 2006;36(7):1379–88.
heavily loaded masonry specimens with the acoustic emission technique. In:
[5] Bazant ZP. Prediction of concrete creep and shrinkage: past, present and future.
7th International seminar on structural analysis of historical constructions:
Nucl Eng Des 2001;203(1):27–38.
Shanghai; 2010. p. 217–222.
[6] Papa E, Taliercio A. A visco-damage model for brittle materials under
[39] Carpinteri A, Lacidogna G. Damage evaluation of three masonry towers by
monotonic and sustained stresses. Int J Numer Anal Meth Geomech
acoustic emission. Eng Struct 2007;29(7):1569–79.
2005;29(3):287–310.
[40] Binda L, Schueremans L, Verstrynge E, Ignoul S, Oliveira DV, Lourenco PB, et al.
[7] Binda L, Anzani A. The time-dependent behaviour of masonry prisms: an
Long term compressive testing of masonry – test procedure and practical
interpretation. Masonry Soc J 1993;11(2):570–87.
experience. In: D’Ayala D, Fodde E, editors. 6th International seminar on
[8] Verstrynge E, Schueremans L, Van Gemert D, Hendriks MAN. Modelling and
structural analysis of historical constructions.: Bath; 2008. p. 1345–1355.
analysis of time-dependent behaviour of historical masonry under high stress
[41] Grosse CU, Ohtsu M, editors. Acoustic emission testing – basics for research –
levels. Eng Struct 2011;33(1):210–7.
applications in civil engineering. Springer; 2008.
[9] Pina-Henriques J, Lourenço PB. Testing and modelling of masonry creep and
[42] Wevers M. Listening to the sound of materials: acoustic emission for the
damage in uniaxial compression. Struct Stud, Repairs Maint Heritage Archit
analysis of material behaviour. NdtE Int 1997;30(2):99–106.
VIII 2003;16:151–60.
[43] BS EN 1330–9 Non-destructive testing. Terminology. Terms used in acoustic
[10] van Zijl GPAG. Computational modelling of masonry creep and shrinkage. PhD
emission testing. British Standard Institution: London; 2009.
thesis. Delft University of technology: Delft; 2000.
[44] BS EN 13477–2 Non-destructive testing. Acoustic emission. Equipment
[11] Choi KK, Lissel SL, Taha MMR. Rheological modelling of masonry creep. Can J
characterisation. Verification of operating characteristic. British Standard
Civil Eng 2007;34(11):1506–17.
Institution: London; 2010.
[12] Ferretti D, Bazant ZP. Stability of ancient masonry towers: Stress redistribution
[45] Ortiz M. Microcrack coalescence and macroscopic crack-growth initiation in
due to drying, carbonation, and creep. Cement Conc Res 2006;36(7):1389–98.
brittle solids. Int J Solids Struct 1988;24(3):231–50.
[13] Roberts TM, Hughes TG, Dandamudi VR, Bell B. Quasi-static and high cycle
[46] Carpinteri A, Scavia C, Yang GP. Microcrack propagation, coalescence and size
fatigue strength of brick masonry. Constr Build Mater 2006;20:603–14.
effects in compression. Eng Fract Mech 1996;54(3):335–47.
[14] Roberts TM, Hughes TG, Dandamudi VR. Progressive damage to masonry arch
[47] McNary WS, Abrams DP. Mechanics of masonry in compression. J Struct Eng-
bridges caused by repeated traffic loading. RCNG 144 Final, report; 2004.
Asce 1985;111(4):857–70.
[15] Abrams DP, Noland JL, Atkinson RH. Response of clay-unit masonry to
[48] Schueremans L. Probabilistic evaluation of structural unreinforced masonry.
repeated compressive forces. In: 7th International brick masonry conference:
PhD thesis. Civil Engineering Department, K.U.Leuven: Leuven; 2001.
melbourne; 1985. p. 567–576.
[49] Casas JR. Reliability-based assessment of masonry arch bridges. Constr Build
[16] Brencich A, Corradi C, Gambarotta L. Eccentrically loaded brickwork:
Mater 2011;25(4):1621–31.
theoretical and experimental results. Eng Struct 2008;30(12):3629–43.
588 A. Tomor, E. Verstrynge / Construction and Building Materials 43 (2013) 575–588
[50] Rüsch H. Researches toward a general flexural theory for structural concrete. J [53] Grgic D, Amitrano D. Creep of a porous rock and associated acoustic emission
Am Concr Inst 1960;57(1):1–28. under different hydrous conditions. J Geophys Res-Solid, Earth
[51] Verstrynge E, Schueremans L, Van Gemert D. Creep and failure prediction of 2009;114(B10201).
Diestian ferruginous sandstone: modelling and repair options. Constr Build [54] Carpinteri A, Lacidogna G, Paggi M. Acoustic emission monitoring and
Mater 2012;29:149–57. numerical modeling of FRP delamination in RC beams with non-rectangular
[52] Lockner D, Byerlee J. Acoustic emission and creep in rock at high confining cross-section. Mater Struct 2007;40(6):553–66.
pressure and differential stress. Bull Seismol Soc Am 1977;67(2):247–58.