Lifetime-Oriented Structural Design Concepts- P8 pptx

crack length during fatigue - plain and circular specimen 3.2.1.1.2 Acoustic Emission The mechanisms by which metals absorb and release strain energy understress, the modelling of which

168 3 Deterioration of Materials and Structures

Fig 3.43 Electrical potential - plain specimen

Fig 3.44 Electrical potential - circular specimen

The distribution of the electrical potential across the plain and circular imen for different crack length is given in Figures 3.43 and 3.44

spec-The results of both the finite element analysis and experiments for bothtypes of specimen are represented in Figure 3.45 as a dimensionless plot of

electrical resistance ratio R/R0 against a/w where R0 is the electrical

resis-tance at the initial crack length a0, R is the electrical resistance at the crack length a and w is half width of the specimen at the crack height A good

agreement between experimental result and the numerical solution of tion of the electrical resistance with the crack propagation can be observed.The obtained evolution of the electrical resistance shows at the same time ahigh level of similarity to the measured crack propagation behaviour undercyclic fatigue load Based on this results, it can be concluded that the method

evolu-of the resistance measurement detects the appearance evolu-of the damage in theearly phase, and it confirms the development of a damage evolution on thebasis of microscopic crack incubation and initiation

Crack length a/w

ANSYS PS

Exper PS

1 1.2 1.4 1.6 1.8

Crack length a/w

ANSYS CS Exper CS

Fig 3.45 Evolution of electrical resistance vs crack length during fatigue - plain

and circular specimen

3.2.1.1.2 Acoustic Emission

The mechanisms by which metals absorb and release strain energy understress, the modelling of which is the basis of fracture mechanics analysis,can be different and complicated Acoustic emission is the elastic energy that

is spontaneously released by materials when they undergo deformation Thestress waves which result from this sudden release of elastic energy due tomicro-fracture events are of most interest to the structural engineer These

events are typically 10μm to 100μm in linear dimension.

Sources of acoustic emission include many different mechanisms of mation and fracture Sources that have been identified in metals include crackgrowth, moving dislocations, slip, twinning, grain boundary sliding and thefracture and decohesion of inclusion

defor-Other mechanisms fall within the definition and are detectable with tic emission equipment These include leaks and cavitation, friction (as inrotating bearings), liquefaction and solidification, solid-solid phase transfor-

acous-mation Sometimes these sources are called secondary sources to distinguish

them from the classic acoustic emission due to mechanical deformation ofstressed materials

Acoustic emission examination is a nondestructive testing method withdemonstrated capabilities for monitoring structural integrity, detecting leaksand incipient failures in mechanical equipment

Acoustic emission differs from most other nondestructive methods in twosignificant respects:

• The detected energy is released from within the test object rather

than being supplied by the nondestructive method, as in ultrasonics orradiography

• The acoustic emission method is capable of detecting the dynamic

pro-cesses associated with the degradation of structural integrity

Acoustic emission expected in fatigue studies is primarily of the burst type.Burst type emission signals originate from sources such as intermittent

170 3 Deterioration of Materials and Structures

Fig 3.46 Definition of simple waveform parameters for a burst-signal

dislocation motion and crack growth in metals A Burst signal, given in theFigure 3.46, has the following characteristic parameters:

• threshold: A preset voltage level that has to be exceeded before an acoustic

emission signal is detected and processed This threshold is independentfor every sensor, and must be chosen depending on the background noise

• burst: A signal whose oscillations have a rapid increase in amplitude from

an initial reference level, followed by a decrease to a value close to theinitial value

• hit: Total signal from the first to the last threshold crossing.

• amplitude: Maximum signal amplitude within duration of the burst.

• duration: The interval between the first and the last time the threshold

was exceeded by the burst

• counts: The number of times the signal amplitude exceeds the preset

threshold

• rise time: The time interval between the first threshold crossing and

max-imum amplitude of the burst

• decay time: The time interval between the maximum amplitude of the

burst and the last threshold crossing

• event: A microstructural displacement that produces elastic waves in

ma-terial under load or stress, which are detected by several AE-transducer.Using time analysis the origin of acoustic emission signal can then bedetected

• event counts: Counts which belong to an event.

3.2 Experiments 171

3.2.1.1.2.1 Location of Acoustic Emission Sources

The ability to locate the sources of acoustic emission is one of the mostimportant functions of the multichannel instrumentation system used in fieldapplication One of the methods for detecting the emission source is the mea-surement of the time differences in reception of the stress waves at a number

of sensors in an array Depending on the sensor location linear (1D), two andthree dimensional problems can be defined

3.2.1.1.2.2 Linear Location of Acoustic Emission Sources

Consider the situation where two sensors are mounted on a linear structure.Assume that an acoustic emission event occurs somewhere on the structure,and that the resulting stress waves propagate in both directions at the samevelocity Using the measurement of the time differences between hits it ispossible to locate position of acoustic emission source If the time differencebetween the hits of both sensors is zero, it would indicate a site preciselymidway between the sensors In general, for the case of constant velocity, thesource location is given by:

d =1

where D is the distance between sensors, V is the constant wave velocity, Δt

is the time deference and d is the distance from the first hit sensor

If the source is outside the sensor array, the time difference measurementcorresponds to the time of flight between outer sensor pair and remainsconstant

3.2.1.1.2.3 Location of Sources in Two Dimensions

The case of location of sources in two dimensions requires a minimum ofthree sensors The input data now include a sequence of three hits and twotime difference measurements (between the first and second hit sensors andthe first and third hit sensors), as can be seen in the Figure (3.47) Then:

Equations (3.8) and (3.9) can be solved simultaneously to provide the location

of a source in two dimensions

172 3 Deterioration of Materials and Structures

Sensor 3

r 1 Sensor 2

X 2 , Y 2 R

or crack initiation, can be observed

The case when acoustic emission is recorded before the previous maximumload is reached is known as felicity effect and describes the breakdown of theKaiser effect If we define the ratio between the load level at which the acousticemission appears and previous maximum load level as felicity ratio, it can beused as the associated quantitative measure of the felicity effect In the case

of the Kaiser effect the value of the felicity ratio is 1

Acoustic emission was detected on two types of specimens The first type

is a plain specimen with thickness of 5mm shown in Figure 3.48 The ond type is a circular specimen with inconstant thickness and outer diameter

Fig 3.48 Geometry of the plain specimen (dimension in mm)

108,169

111,883 146 150 166

Fig 3.49 Geometry of the circular specimen (dimension in mm)

166mm shown in Figure 3.49 All specimens were made from heat-treatablesteel 42CrMo4 (No 1.7225)

Two types of AE-transducers with appropriate preamplifiers were used forthe detection of acoustic emission The first set was 4 piezoelectric transducersR15 with resonant frequency 150kHz and 4 single In-Line 40dB preamplifierswith a 100-300kHz bandpass filter The second set was 4 wideband piezoelec-tric transducers WD with operating frequency range 100-1000kHz and 4 volt-age preamplifiers with 20/40/60dB selectable gain and 100-1200kHz bandpassfilter All acoustic emission signals were amplified with 40dB and recorded us-ing Physical Acoustics software on a two Two-Chanel-AE-Boards The samplerate for all measurements was 10MHz All AE-transducers were clamped tothe specimen with a spring clamp The coupling between the specimen andthe transducers was made with a silicon gel

The acoustic emission threshold was set to 35dB, as a compromise betweeneffectively avoiding background noise and cutting off low level signals dur-ing damage evolution The threshold setting depended on the experimentalconditions

174 3 Deterioration of Materials and Structures

40

x 117

Fig 3.50 The position of AE-transducers on the plain specimen

98

4

2 1

3

Fig 3.51 The position of AE-transducers on the circular specimen

The position of AE-transducers on the plain and circular specimen is given

in Figures 3.50 and 3.51 respectively The position of AE-transducers on theplain specimen was given in such a way, that transducers 1 and 4 are so-called guard transducers with the function to eliminate signals originatingoutside the specimen test section from the recorded data Thus, extraneoussignals such as those emanating from load-chain noise or from servo-valvesand hydraulic pump were avoided without loss of data The AE-transducer 2and 3 were used as measuring sensors In the case of the circular specimen all

4 transducers are measuring sensors

3.2.1.1.2.6 Experimental Results

Acoustic emission recorded during the test is represented using acousticemission events counts per cycle over the whole stress range and cumulativeacoustic emission event counts during fatigue damage For all experiments, theload was in the range, which leads to high cycle fatigue with brittle damage.This type of load and the brittle damage behaviour lead to the significant

Fig 3.53 Acoustic emission event count rate during fatigue - circular specimen

increase in the acoustic emission output, as the crack advances towards finalfailure

The rate of acoustic emission in the form of acoustic emission events countsper cycle over the whole stress range is given in the Figures 3.52 and 3.53.When the load was in the elastic range, the low acoustic emission output wasevident during initial cycles, due mostly to microscopic dislocation dynamics.This stage was followed by a dead period with almost no acoustic emission Inthis stage of fatigue damage accumulation results in the long crack-initiationlifetime Low energy dislocation motion, which generates acoustic emission

176 3 Deterioration of Materials and Structures

Fig 3.54 Acoustic emission total event counts during fatigue - plain specimen

Fig 3.55 Acoustic emission total event counts during fatigue - circular specimen

waves, is frequently under background noise and is relatively hard to tect Only discrete acoustic emission events counts represent the existence ofacoustic emission and consequential evolution of fatigue damage The crackpropagation, which is connected with the high rate of the acoustic emission,occurs in the third stage The release of the elastic energy due to the crackpropagation has a significant level and the detection of the acoustic emission

de-is not influenced by the background node-ise as in the second stage of the fatigue.Evolution of cumulative acoustic emission event counts during fatigue dam-age is given in the Figures 3.54 and 3.55 and represent a cumulative fatigue

168 128 60

168 128 60

0

Location [mm]

PS31

Fig 3.56 The location of the origin of acoustic emission for the plain specimen

Fig 3.57 The location of the origin of acoustic emission for the circular specimen

damage process This process can be divided into the same three stages asindicated for the rate of acoustic emission: (i) microscopic dislocation dynam-ics; (ii) microscopic crack incubation and initiation; (iii) macroscopic crackpropagation

The location of the origin of acoustic emission was computed using timedifference measurement methods described earlier The obtained distance rep-resents the distance between origin of acoustic emission and AE-transducers.The location of the origin of acoustic emission for the plain specimen is given

178 3 Deterioration of Materials and Structures

60 30

120 90 60 30 0

120 90

60 30 0

Amplitude [dB]

CS02

0 125000 250000 375000 500000

120 90 60 30 0

Amplitude [dB]

CS05

Fig 3.59 Acoustic emission event counts against amplitude for the circular

speci-men

in the Figure 3.56 and for the circular specimen in Figure 3.57 As can be seen

in the Figures, the computed location of the origin is in good agreement withthe real position of the fatigue damage Most events are located at the positionwhere the crack is present, confirming that the acoustic emission is from thecrack As a consequence of the specimen size, there are inherent problems withthe location due to the size of the AE-transducers compared to their spacing.Namely, spacing between AE-transducers was 68mm for the plain specimenand between 90-98mm for the circular specimen and the diameter of trans-ducers is 18mm, which excludes a point representation of AE-transducers Forthis reason, signals which are located within 10mm of the real crack positionare regarded as well located

Acoustic emission event counts with respect to the amplitude are given

in the Figures 3.58 and 3.59 As can be seen most of acoustic emission

Frequency [kHz]

PS31

Fig 3.60 Acoustic emission event counts against frequency for the plain specimen

event counts occur with the amplitude between 40 and 70dB This behaviourcould be used as an additional condition in the elimination of the extraneousnoise

Using wideband AE-transducers gives a possibility to investigate the quency at which the acoustic emission occurs In the Figures 3.60 and 3.61acoustic emission event counts versus the frequency are given Since al-most all acoustic emission counts have a frequency between 50-300kHz, res-onant piezoelectric transducers R15 with resonant frequency 150kHz can

180 3 Deterioration of Materials and Structures

3.2.1.2 Degradation of Concrete Subjected to Cyclic Compressive

Loading

A great number of concrete structures is exposed to cyclic mechanical ing scenarios Therefore, the reliability of such structures depends among otherinfluences also on the degree of structural degradation due to fatigue loading

load-In order to estimate the state of a structure it is necessary to know the velopment of the degradation of the material properties during its lifetime.However, up to now a statistically based description of the degradation pro-cesses and their effects on the compressive strength, stiffness and fractureenergy referring to pure cyclic compression loads of plain concrete are stillmissing

de-For this purpose, within the large joint research project (Collaborative search Center 398) extensive experimental investigations were carried out atthe Ruhr-University in Bochum In order to get information about the degra-dation processes with sufficient reliability a large number of specimens weretested by measuring ultrasonic transmission time, longitudinal strains andstress-strain curves during cycling loadings [148, 149, 402]

Re-3.2.1.2.1 Test Series and Experimental Strategy

Most of the extensive cyclic tests were performed on normal-weight concrete

of grade C 30/37 Furthermore also high strength concrete of grade C 70/85

as well as air-entrained concrete (grade C 30/37) were investigated Withinthe C 30/37 series the types of the coarse aggregates (quartzite, basalt andsandstone) were varied Additionally concretes with different coarse gradingcurves were tested, whereby in these cases the matrix of mortar was keptconstant

For all the tests cylindrical specimens with a diameter d of 100 mm and aheight h of 350 mm were used These specimens were taken as cores, drilledfrom concrete blocks at an age of about 25 days In comparison to specimensmade in separate formworks, the drilled specimens have no accumulations offine-grained mortar along the surface and represent a part of a real concretestructure in a better way Thereby additional impairment of the peripheralzone can be prevented

The cyclic tests normally started at a concrete age of about 40 days vious the specimens remained on air; within these about two weeks betweencore-drilling and test-start nearly constant hygral conditions in the concretespecimens (equilibrium moisture content) could be reached Most of the cyclictests were performed as single-stage tests at constant stress levels In some fewseries also two-stage tests with an alternating upper stress level were carriedout For all tests a hydraulic cylinder pulsators were used; the frequency ofthe cyclic loading was constantly f = 7 Hz

Revealing the degradation process

Fig 3.62 Execution of the single-stage and two-stage test

In the single-stage test series the specimens were loaded within a defined

stress-range Smax/Smin(Figure 3.62, left) In all of these tests the lower stress

level Smin was adjusted constantly at 0.10 (i.e 10 per cent of f c), while the

upper stress level Smaxwas varied from series to series between 0.75, 0.675 and

0.60, however, within one series the Smaxwas kept also constant Furthermoreinterruptions within these single-stage tests were investigated, mainly with theaim to check, to which extent degradations can recover in such rest periodsbetween various load scenarios

Concerning the two-stage tests the lower stress level was also kept constant

at Smin = 0.10, whereas the upper stress levels were varied within one seriesafter a defined number of load cycles (Figure 3.62, right) In these series

various sequences for the upper stress level were considered (Smax = 0.60 →

Smax = 0.675 and v.v.; Smax = 0.75 → Smax = 0.675 and v.v.) So in both cases the tests were started with a higher Smax and then reduced to lower one

as well as they started at a lower Smax and afterwards the upper stress levelwas increased

During these cyclic tests the longitudinal strains were measured ously by two strain gauges (50 mm in length) which were applied in axialdirection on opposite sides of each specimen

continu-Additionally microdefects and their development were investigated by destructive ultrasonic (US) measurements perpendicular to the main direction

non-of stress after applying a defined number non-of load cycles Furthermore at thesame stages also the static Young’s modulus was determined, however, in this

182 3 Deterioration of Materials and Structures

case in the main load-direction, by carrying out an additional load step at a

Smax/Smin -ratio of 0.30 / 0.10 during a short interruption of the cyclic test

In addition to these non-destructive tests, the development of the ical properties - especially the changes in the stress-strain curve and strength

mechan were investigated by destructive tests on specimens taken out of the setupafter certain defined numbers of cycles By comparison of the so obtained vari-ous stress-strain curves the development of the strength, stiffness and fractureenergy of the respective concrete could be described

3.2.1.2.2 Degradation Determined by Decrease of Stiffness

The proceeding in the degradation process is mainly described by the changes

in the stiffness This can be demonstrated also within the performed

test-series by means of the Young’s modulus Estat (Figure 3.63) as well as thedynamic E-modulus (3.64) For all considered stress levels a typical sharpdecrease could be observed after only a few number of cycles Exemplarily at

the stress regime Smax/Smin = 0.675/0.10 the Young’s modulus Estat as well

as the dynamic elastic modulus Edyn decreased within the first 180,000 loadcycles by about 12.5 per cent (averaged value) A following steady decline with

a significant lower slope was – also typically – observed between 180,000 and400,000 cycles After applying about 400,000 cycles some specimens showed anaccelerated decrease in these characteristic values Additionally an increasing

scatter of the measured Estat and Edyn by increasing number of cycles was

determined So, e.g., the coefficient of variation in the Edyngrows up from 40.1per cent after 10,000 cycles to 79.8 per cent after 600,000 cycles Especiallyafter about 400,000 cycles a significant increment in the standard deviationwas observed

Fig 3.63 Decrease and scatter ofEstat at Smax/Smin = 0.675/0.10

(single-state-tests)