1:
2:
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: %\documentstyle[pdftex,aps,epsfig,multicol,12pt]{revtex}
5: %\setlength{\textheight}{230mm}
6: %\documentstyle[preprint,epsfig,aps]{revtex}
7: %\draft
8: \documentclass[aps,multicol,psfig,epsf,epsfig,preprint,showpacs]{revtex4}
9: % \documentclass[pdftex,aps,prl,twocolumn,showpacs]{revtex4}
10: \usepackage{graphicx}
11: \usepackage{amssymb}
12: \usepackage{amsmath}
13:
14: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
15:
16:
17:
18: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
19:
20: \begin{document}
21:
22: \title{Crackling noise in paper peeling}
23:
24: \author{L.I. Salminen$^1$, J.M. Pulakka$^1$, J. Rosti$^1$,
25: M.J. Alava$^{1,2}$, and K.J. Niskanen$^3$}
26:
27: \affiliation{$^1$Helsinki University of Technology, Laboratory of Physics,\\
28: P.O.Box 1100, FIN-02015 HUT, Finland }
29:
30: \affiliation{$^2$SMC-INFM, Dipartimento di Fisica,
31: Universit\`a "La Sapienza", P.le A. Moro 2
32: 00185 Roma, Italy}
33:
34: \affiliation{$^3$Finnish Pulp and Paper Research Institute, P.O.Box 51,
35: FIN-02151 Espoo, Finland}
36:
37: % tiiviste
38: \begin{abstract}
39: Acoustic emission or crackling noise is measured from an experiment
40: on splitting or peeling of paper. The energy of the events
41: follows a power-law,
42: with an exponent $\beta \sim 1.8\pm 0.2$. The event intervals
43: have a wide range, but superposed on scale-free statistics
44: there is a time-scale, related to the typical
45: spatial scale of the microstructure (a bond between two
46: fibers). Since the peeling takes place via
47: steady-state crack propagation, correlations can be
48: studied with ease and shown to exist in the series of acoustic events.
49: \end{abstract}
50: \pacs{62.20.Mk,05.40.-a, 81.40Np}
51: \maketitle
52: \date{\today}
53:
54: % intro
55: There are several experimental signatures of power-law statistics,
56: or scale-invariance, in fracture. One interesting case
57: is acoustic emission (AE),
58: which is produced by the release of small quantities of elastic
59: energy during the failure of an inhomogeneous sample, and is an
60: example of the phenomenon of ``crackling noise'', met in many
61: kinds of systems in physics \cite{sethrev,Bourev}. This
62: can result in analogies of the
63: Gutenberg-Richter's and Omori's laws for earthquakes
64: \cite{Ciliberto,penn,Petri2,Lockner,oma}.
65: These are experimental observations; the former relates the probability
66: (for earthquakes or AE) of an event $P(E)$
67: to the energy or magnitude of the event $E$ by an exponent $\beta$.
68: Correspondingly, the latter binds by an exponent $\alpha$ the waiting time
69: $\tau$ between the events and their probability $P(\tau)$. Very often
70: these laws are witnessed simultaneously in both microfracturing
71: and earthquakes \cite{simu}. The origins of both the
72: power-laws are mostly unknown and the same holds also for
73: the possible connection between these.
74: %For earthquakes, the exponent
75: %values are typically $\alpha \approx 1.0 $ and $\beta \approx 2.0$ \cite{gr}.
76: %{\bf tama referenssi ei liene kovin hyva...}.
77: % lisa-intro
78:
79: The common idea about material failure is based on a critical
80: defect: the sample fails catastrophically once the local strength,
81: that follows from among others the size of the flaw, is exceeded.
82: The observation of AE implies often that microscopic damage
83: is being created, contrary to one, single catasthropic
84: crack growth event. One consequence is that, in analogy
85: to earthquakes again, the idea of predictablity becomes
86: of interest. This is due to the precursors to final
87: failure that could be diagnosed by AE. In general terms
88: it is of fundamental materials science and statistical
89: physics interest that universal or critical fracture behavior should
90: exist in the presence of varying material properties, like anisotropy,
91: related to the shape of the stress-field close to microcracks or
92: a notch, or to e.g. asymmetric disorder \cite{lopinaa}.
93: Any power-laws in AE or crackling noise should originate
94: from a basic, fundamental mechanism, in spite of such complications.
95:
96: One particulary inviting proposal
97: is to follow the passage of a crack front through a quasi
98: two-dimensional sample, which can be realized eg. in a
99: weak interface between two three-dimensional elastic plates \cite{line,Fisher}.
100: This scenario has the advantage, over say ordinary tensile
101: tests, that the crack propagation takes
102: place in a {\em steady-state} in contrast to most other experiments
103: of fracture. In plexiglass, the front may be self-affine,
104: and a roughness exponent $\chi$ close to 0.6 has been measured,
105: which is intriguing given that the advancement takes
106: place in avalanche-like events \cite{maloy}. The theoretical understanding
107: of the phenomenon is based on numerical models and on stochastic
108: equations for the crackline dynamics. These have so far not
109: been able to account for most of the observed features, nor
110: is there any understanding of whether the roughening could
111: be expected to be ``universal''.
112:
113: In this work we look at the statistics of fracture AE
114: in a set-up that mimicks such line dynamics in ordinary paper
115: by peeling a sheet into two, as a crack advances through it
116: (Figs. \ref{figyksi} and \ref{figkaksib}).
117: We obtain that the energy of AE events scales as a power-law,
118: like in tensile tests, which produce analogies
119: of both the Gutenberg-Richter and Omori's laws.
120: This takes place with
121: a novel exponent which is much larger than those usually
122: met in fracture AE. The temporal statistics exhibit
123: a wide spectrum of intervals between the events. Translated
124: into the average distance that the fracture line propagates
125: we show below that there is a typical scale, related to
126: the size of a fiber-to-fiber bond.
127: Rupture processes often involve time-dependence, but at least
128: for low strain rates the structural scale stays the same.
129: The dynamics of the peeling line is most likely
130: quasi-one-dimensional. That is, due to the thinness of a
131: sheet of a paper the stress field is not able to penetrate
132: into the intact paper so as to give rise to two ``scalefree''
133: lengthscales \cite{zapperi}. We also compare to other kind of AE data
134: for the same materials, demonstrating that for localized
135: fracture, the $\beta$-exponent is much larger, and that
136: in a tensile test values met before can be obtained \cite{oma}.
137: Such experimental data should be useful to
138: to formulate and test microscopic fracture models.
139:
140: In the case of paper, Yamauchi made pioneering acoustic emission
141: measurements \cite{yama}, with the claim that fiber and bond
142: breakages can be distinguished in the amplitude histogram
143: of the signal.
144: The idea of basic scales
145: should be contrasted with Fig.~\ref{figkaksib} that introduces
146: the crack advancement scenario. Due to the nature
147: of the set-up, the stress-field
148: is expected to be cut-off quickly (in some models exponentially)
149: with increasing distance from the average crackline position.
150: Thus the individual
151: events, interpreted in terms of the area over which the ``avalanche''
152: passes, are in practice one-dimensional. Of course, the crack can
153: fluctuate on very small scales in the $z$-direction, perpendicular
154: to the sheet.
155: % The
156: %typical area of a fiber-to-fiber bond can also be related to the
157: %timescale via the strain rate, and the advancement of the fracture
158: %line will for sure influce also the breaking of fiber-to-fiber
159: %bonds breaking, whether partly or as part of larger events.
160:
161: With intact samples in a tensile test the strength
162: depends on weak spot statistics. In practice some existing weak region
163: often launches the crack growth, and after that rupture becomes mainly
164: a local problem. Naturally, if a sizeable initial (edge or center)
165: notch is applied, the fracture process focuses in the
166: so-called fracture process zone, FPZ,
167: around the crack-tip (for paper, the characteristic dimension of
168: FPZ is upto 5 mm, typically).
169: Since fracture can happen in a diffuse or in such
170: a localized way one might see two separate statistics related to
171: these processes, and distinguish two regimes. These would be the
172: pre-fracture phase, when behavior is nearly elastic and cracking or
173: damage disperse and the second part taking place after and at the
174: stress maximum. During this regime a single crack is propagating and
175: the failures concentrate in the FPZ, the vicinity of the crack-tip.
176:
177:
178: In the presence of disorder the fracture can be an irregular process of
179: elementary rupture events separated by interarrival or waiting times,
180: and spread out geometrically in the sample
181: The variety of theoretical or computer models available
182: differ in the level and details of load resharing after microfailures
183: \cite{Zapperi}.
184: The events in such studies consist of
185: single spring-like breakages, with a varying number of those elementary
186: breakages within one coherent event. In the
187: case the fracture takes place via the avalanche-like dynamics,
188: 2d computer simulations indicate a $\beta$-value of 1.7 \cite{recent}.
189: Also, $\alpha = 0.94 \pm 0.20$ was reported in a simulation of hydraulic
190: fracture \cite{herrmann}. All such studies do not apply to the peel-experiment
191: at hand, ie. there are no theoretical results pertaining to the
192: possible AE statistics in our case.
193:
194: %The main conclusion is that the dynamics gives
195: %rise to a scale-invariant distribution of event (``crackling
196: %noise'') energies, with an exponent $\gamma$ that differs
197: %from the one that can be observed in tensile tests. Meanwhile
198: %there is clear evidence that though the microscopic timescales
199: %have also a very wide distribution, there is also a characteristic
200: %time that maybe relates to the particular nature of the material.
201: % ae:sta
202: Acoustic emission as such is a well-known technique to monitor fracture in
203: e.g. composites. The rapid release of elastic energy can be observed by
204: ultrasonic sensors \cite{corte}, with little influence on the actual
205: fracture process.
206: %We are in the following interested in generic
207: %observations on fracture, based on the AE, while
208: %the precise origin of acoustic events is usually unknown.
209: % lupailua
210: %The setup presented below gives insight into the dynamics
211: %of fracture in disordered media, by also considering the time-dependence
212: %via varying strain rates and by comparing to
213: %the results of tensile tests on
214: %exactly the same material both with and without a notch.
215: %It also presents steady-state
216: %experimental conditions, as a crackline propagates through
217: %the sample.
218: % dennis rautalanka
219: The peel-in-nip method to split paper is based on the
220: nip between two rolls rotating synchronously (Fig. \ref{figyksi}).
221: The front end of the sample is attached to both the
222: rolls by tape and then the cylinders are reeled to initiate the peeling.
223: Then the cleavage proceeds based on an equilibrium of three supporting
224: forces. The peeling takes place near the nip so that angle between the
225: planes of the halves is close to $\pi$ and the actual value
226: has a slight dependence on the particular kind of paper.
227: The test produces very large fracture surfaces; over 100 mm${}^2$ compared to
228: a typical area 1 mm${}^2$ in standard tensile tests.
229: Paper is a fibrous material, such that the fibers
230: form interpenetrating layers.
231: In standard tests the fibers are loaded mostly in-plane,
232: and several microscopic failure mechanisms co-exist (bond
233: or fiber breakages, fiber pull-out).
234: The fractureline separating the intact part and the separated
235: halves is continuous, however the paper structure is discrete on
236: scales below the fiber length. Thus some bond dimension related
237: cross-over in any distribution (energy or intervals of events, or durations)
238: is possible. Figure \ref{figkaksic} demonstrates a typical
239: test: note both that there is a maximum scale for interarrival
240: times, and that, of course, the AE event size has no trend during the test.
241:
242: % naeytteet
243: We used two sets of handsheets, paper made in a standard laboratory
244: mould from standard mechanical pulp. Typical tensile strength is
245: 3 $kN/m$ and strain at break $2 \%$.
246: One control parameter used is refining, or beating the pulp
247: which makes the fibers more flexible and more fibrillated.
248: Due to that the resulting paper is more uniform and becomes thus stronger.
249: Contrary to industrial paper the fiber orientation is uniform.
250: We had a set of handsheets with six different refining levels.
251: %To test
252: %the influence of fiber breakages, we also had sheets with varying
253: %level of viscose fibers. The viscose fibers are made of cellulose too,
254: %but they are much stronger and their bonding capability is weaker than
255: %it is for the wood fibers.
256: Small (length 70 mm, width 15 mm) sample strips were used to reduce
257: the elastic energy often leading on catastrophic crack growth,
258: in ordinary tensile laboratory tests performed for comparisons.
259:
260: % tensilenotch
261: %Tensile test is very common method to evaluate strength of material.
262: %The existence of a notch is experimentally of great concern. For weak
263: %materials (wood, plastic) notches are seldom used and for strong
264: %materials (steel) notch is essential. In the case of paper, most of
265: %experiments are still done without a notch. Nowadays the usage of
266: %notched testing is increasing.
267:
268:
269: % ae suureet
270: During the experiment we acquire bi-polar acoustic amplitudes
271: simultaneously on two channels by piezocrystal sensors
272: (Physical Acoustic Corporation R15 tranducers, resonant
273: frequency 150 kHz)
274: as a function of time. In addition the force
275: of peel-in-nip or tensile test was measured, typical values being
276: a few tens of Newtons depending on the thickness and rolling
277: friction of the grade of paper tested (in Fig. \ref{figkaksic}
278: about 90 N, with a variability of about 10 N) .
279: In tensile tests the two transducers were attached directly to paper and
280: no coupling agent was used. Each channel has 12-bit resolution and
281: a sampling rate of 400 000/s. The transmission time from event origin to
282: sensors is order of 5 $\mu$s. The acoustic channels were first
283: fed through custom-made amplifiers,
284: and after that held using sample-and-hold circuit. The
285: shape of AE pulses can change and attenuate during transmission,
286: but the effect should have has only minor effect to our analysis.
287: %The sample-and-hold circuit is need to cut off the 2.5 $\mu$s
288: %time-lag between ae-channels.
289: The acoustic time-series are reformed
290: offline by thresholding, by the detection of continuous and coherent events,
291: and by the calculation of event energy $E$, the sum of
292: squared amplitudes within the event. Events are separated by silent
293: (i.e. amplitude below threshold level) waiting intervals $\tau$. We do
294: not interpret the waveforms of the events but analyse the data in
295: the statistical sense. The energy span is estimated to be
296: about 1 $\mu$J to 1 mJ; recall that the energy released
297: ends up as heat eventually.
298: In general the energy of the event is expected to be
299: proportional to the damaged area corresponding to the event
300: \cite{lys} and to the stress in that area.
301: %The amplitude attenuation is
302: %close to a $r^{-2}$ law.
303:
304: % gr-perustuloksia
305: The energy statistics shows that
306: acoustic events obey a Gutenberg-Richter -like
307: power-law for the peel-in-nip test (figure \ref{figkolme})
308: with the value $\beta=1.8 \pm 0.2$.
309: The figure provides as a comparison the energy data from
310: an ordinary tensile test: the exponent $\beta=1.2\pm0.1$, in
311: reasonable agreement with earlier such experimental values.
312: These can also be compared to the value
313: from a tensile test with a {\em large} initial notch,
314: resulting in the exponent $\beta = 1.7\pm0.2$. Note that
315: the peel-in-nip case presents some slight curvature.
316: The difference in the $\beta$-exponent between this
317: data and the tensile-with-notch appears us to be
318: statistically reliable, in spite of the error bars.
319: Notice that as is usual in an AE test the energy scale
320: can not be calibrated quantitatively.
321:
322: % gr-spekulaa
323: The different $\beta$ -values imply that there are scale-free
324: behaviors in both tensile and peeling fracture, in the same
325: material. Distributed damage all over the sample and microcrack
326: coalescence, as in a
327: tensile test, produces a small value for $\beta$.
328: Such failures take place at weak spots with high local
329: stresses. Any microscopic breaking stresses (fibers, bonds) will
330: have bounded distributions (e.g. Gaussian), so the
331: power-law distribution of energy implies that
332: the number of elements (i.e. size) of rupture
333: events has to vary in power-law manner. In the peel-in nip test
334: (and also with tensile samples with a large notch)
335: the process reflects a more localized failure, but still scales as
336: a power-law albeit with a much higher exponent.
337: In the peeling experiment, most likely the dynamics
338: is one-dimensional - the essential variation in the area covered by
339: events, corresponding to the energy takes place along the crack
340: line but not into the sheet, in the planar direction.
341: %In the end of tensile test failure can happen also in
342: %localized way and thus the localized fracture power-law is present.
343: %But since the disorder induced exponent of diffuse rupture is lower
344: %(1.2) we see no sings of bond breakage exponent $\beta$ = 1.7.
345: %Higher
346: %exponent value denotes of more homogeneous rupture scheme, which in
347: %general are known to result higher exponents than disordered rupture.
348: %Attenuation is not plausible as a reason for the exponent value
349: %difference, but note that the range of
350: %energy differ sin tensile and peel-in-nip tests (figure
351: %\ref{figkolme}). This is due to the differing sensor attachment.
352:
353: % locknerin nature b-kamaa
354: The variation in the exponents is analogous to the work by
355: Lockner et al., on fracture and crack growth in
356: granite \cite{Lockner}. They observed
357: energy scalings with $\beta = 1.2 -- 2.3$, due to a setup
358: that allowed to follow the variation of $\beta$ during the fracture.
359: In that context it
360: was suggested already earlier, by a model, that
361: $\beta$ should drop at maximum stress \cite{week}.
362: In addition the exponent was expected to recover after
363: the start of crack propagation, as was seen qualitatively
364: in the experiments of Lockner et al:
365: $\beta$ dropped from a value above 2 to near 1.2 at maximum
366: stress, and recovered back to a value of 1.7.
367: % This behavior is in nice
368: %agreement with our results if we consider all tensile events to
369: %results from local maximum stress ruptures, which initiate new
370: %cracks.
371: %The argument is indeed reasonable, since it's common belief
372: %that before the catastrophic failure of paper the breakages originate
373: %from local stress pinnacles.
374: %In the work by Lockner et al. the change in statistics was combined
375: %to a rough picture of crack localization.
376: % studied also the acoustic emission source location and
377: %somewhat it succeeded. But note that
378: %To gain any true information
379: %about the proceeding of the fracture process, the localization should
380: %have better resolution than the smallest heterogeneity in sample. In
381: %case of paper that limit is order of 100 $\mu$m corresponding to time
382: %resolution of 50 ns.
383: %In literature the highest localization accuracy
384: %reported is $\pm$3 mm by Lockner et al (cite lockner).
385: % \footnote{en ole $\beta$:sta varma} Note that by using pulp
386: %refining one can induce slight variations to $\beta$, between
387: %1.1 $\to$ 1.4, in the tensile comparisons. If taken literally
388: %this would imply, that the exponent can be changed continuosly
389: %due to the influence of the material properties.
390: %(cite
391: %J. Weiss, F. Lahaie , J.P. Grasso, Journal of Geophysical Research,
392: %Vol 105, No B1, pp 433-442, 2000). They also proposed critical points,
393: %disordered first-order transitions and self-organized criticality as
394: %candidates to produce power-law statistics.
395:
396: % jauhatussarja
397: %For pulp refining serie we observed that beating has slight effect on
398: %Gutenberg-Richter's law's exponent. For unbeaten pulp, the value of
399: %$\beta$ is close 1.4, for moderate beating about 1.2 and for very
400: %heavy beating near 1.1 \footnote{ehka noita jauhatus exponentteja ei
401: %tarvitsisi huudella julkisesti ollenkaan, kun ne on vahan
402: %epavarmoja}. In general unbeaten pulp gives more events, but their
403: %distribution is relatively narrow. Refining seems to flatten out the
404: %energy distribution (figure \ref{figkolmeb}). In addition larger
405: %events are present is beaten pulp rupture than in unbeaten one. This
406: %result is quite strange, since we can alter the power-law exponent by
407: %parameter change. Possibly explanation of the effect of the beating
408: %is the drastic change of network geometry and thus also the fracture
409: %mechanism changes.
410:
411:
412: % omo-perustulosta
413: The waiting time distribution in tensile tests often results in a distinct
414: power-law \cite{oma}. With a large notch there is some
415: scattering from pure scaling behavior. In both cases the exponent
416: $\alpha$ is close to 1.0 (figure \ref{fignelja}). The peel-in-nip
417: experiment diverges clearly from such an ideal dependence in the
418: time slot 10 - 500 ms. The fracturing takes place in a steady-state,
419: but such that maximum waiting times take into account that
420: the crack line has minimum velocity, imposed by the angular
421: velocity of the rolls.
422: % omo-laetinaeae
423: The $\alpha$ values reported in literature are without exception in the
424: proximity of unity. For instance,
425: in a creep experiment with
426: a cellular glass material Omori's and Gutenberg-Richter laws were
427: observed, with the values for
428: $\alpha=1.3$ and $\beta=1.5$ \cite{Maas}.
429: %In addition Meas
430: %et al. note the that distance between consecutive events shows also scale
431: %invariance $P(d) \sim d^{-1.6}$.
432: In AE experiments with ice $\alpha=1.0
433: \pm 0.3$ and $\beta = 1.3$ \cite{geofyysikoita}. These should be different
434: from the peeling
435: set-up, where the events by force take place at or very close to the
436: crack line.
437: Given $\alpha \sim 1$ the mechanism producing the scaling in all
438: these might presumably be universal.
439: %Since the waiting time usage is
440: %equal for all waiting times ($\tau P(\tau)
441: %\approx $const.) when Omori's exponent is near 1, we emphasize the
442: %contradiction between the unstationary nature of tensile experiment
443: %and ``constant rupture velocity''% plead by waiting times.
444: %To our
445: %knowledge theoretical proposals to explain this disagreement are
446: %missing.
447:
448:
449: % gr-omo-hopinaa
450:
451: % visgoose
452: %Viscose fibers do not break and bond (HETKO:together or with wood ?)
453: %poorly, so adding viscose emphasize the bond-ruptures (left figure on
454: %\ref{figviisi}). The slope of Gutenberg-Richter power-law is modified
455: %with viscose level. It is very seldom that the magnitude of
456: %constituent attribute affect the scaling exponent. In tensile test the
457: %energy distribution is wider. In localized fracture in peel-in-nip
458: %test the energy scaling maintains invariable (right figure on
459: %\ref{figviisi}). Since peel-in-nip breaks bonds only (so $\beta=1.7$ is
460: %bond-breaking), viscose level has no additional effect.
461:
462: % strain-rate asja denniksessa
463: In order to study the origin of the typical scale
464: in waiting time distribution,
465: we did peel-in-nip experiments with various strain rates. We observed
466: that the position of the ``leap length'' plateau shifts
467: %(figure\ref{figkuusi})
468: such that the lengthscale, interpreted as the distance
469: that the fracture line advances in a time corresponding to
470: the position of the plateau, stays roughly constant. It
471: is located between 1 $\mu$m and 50 $\mu$m, corresponding to a typical
472: fiber width and fiber-to-fiber bond scale in paper.
473: For a strain rate (500mm/min) the leap plateau
474: disappears. This may originate from strain at the wire pulling
475: the rolls or inertia of the rolls, or the elastic energy stored into
476: strained part of the sample.
477: Without scaling the initial and final
478: parts of the distributions overlap, which
479: implies the presence of other effects related to the fast
480: and slow timescales.
481:
482: It is interesting to note that the Gutenberg-Richter exponents are close
483: in tensile tests with notch and peel-in-nip tests and
484: that the Omori exponent
485: has close value in tensile and tensile with notch tests.
486: These experimental findings underline that these two power-laws are
487: produced by
488: separate mechanisms. In particular it is possible to distort the
489: energy scaling, only.
490:
491: % voima-haessaekkae
492: %{\it Mun tai joonaksen pitäisi vähän katsella mitä se voima denniskokeessa
493: %oikein tekee.} The reeling moment has large variations during the
494: %peel-in-nip test. The question thus is if the variations are in
495: %correlation with acoustic emission. Our hypothesis is that at or
496: %immediately after an event the moment should diminish. Then the moment
497: %should recover some fraction of its value before the event. During
498: %stable stage the moment might be noise, since the apparent paper thickness
499: %is not a constant.
500:
501: % potenssilaetinaeae
502: %The experiments show evidence of two distinct rupture schemes. The
503: %dominating one is the disperse failure which is consequence of
504: %disorder in material. The second mechanism is always present too and
505: %affected by network structure, segment activation, fiber and bond
506: %strengths. This result explains why the strength of elements is not
507: %responsible of paper strength in standard test. For reason that still
508: %remains unknown the rupture in tensile test takes place almost only in
509: %the local stress maximum manner. In peel-in-nip test stress maximum is
510: %not the main rupture factor, but the fracture is initiated by some
511: %other mechanisms. \footnote{Honestly i cant understand at all how the rupture
512: %could originate from anything else except stress maximum}. Maybe the
513: %phenomenon could be explain by saying that local strain which is not
514: %result of global stress but, local displacement causes the failure. So
515: %in diffuse rupture case the reason is global stress and the exponent
516: %$\beta = 1.2$ . On the other hand in local rupture the reason is local
517: %strain and the characteristic exponent is $1.7$.
518:
519: For short enough time-windows the fracture
520: line may advance in a correlated fashion. This is demonstrated in
521: Fig.~\ref{figseit}. One observes that large events, in terms
522: of energy, are inter-correlated and more likely to follow each
523: other than if the signal was completely random. The natural
524: interpretation is that stronger crackline pinning is overcome
525: in a {\em correlated} fashion, so that the consecutive events
526: bear signatures of the energy stored due to the constant load
527: rate, during periods of no acoustic activity. One can also study
528: the autocorrelation functions (since the signal is stationary).
529: This results in a similar picture, in that the signal is correlated
530: upto a timespan which reaches about 0.05 s (depending on the straining
531: rate), as depicted in Fig.~\ref{acf}. Note that for the slower
532: rate the correlations extend further. The reference cases are
533: artificial signals using a randomized sequence of (exactly the
534: same) events and intervals.
535: We also attempted to correlate the AE data with the force
536: signal of the tensile testing machine; unfortunately it
537: became apparent that the effective force at the crackline
538: is masked by the two separate sheethalves that transmit
539: it from the rolls. In particular no direct correlated
540: between $E(t)$ and $F(t)$ could be detected.
541: The timeseries as such exhibited $1/f^\alpha$ noise, with
542: an exponent close but below unity.
543:
544: % holynpolya
545: To conclude we have made experimental observations of acoustic
546: emission in paper peeling. The set-up is such that steady-state
547: crack propagation across a quasi-two dimensional material can
548: be followed. The main discoveries are a power-law for the
549: energies of the acoustic events, in spite of the fact that
550: the process is confined to an almost one-dimensional geometry,
551: and for the event waiting time statistics. The former supports some
552: earlier observations, together with the tensile tests, that
553: a strong localization of the fracture process produces scalefree
554: statistics with a large exponent. The latter shows an
555: intriguing deviation from a power-law, which seemingly can be
556: attributed to the microstructure of paper.
557:
558: %: bonds between fibers
559: %give a length-scale and a time-scale.
560:
561: %which
562: %support the scheme proposed by Lockner. We also are able to explain to
563: %consequences of two fracture mechanisms.
564: %In our opinion the nominal strength of paper is reflection of local
565: %disorder. The disorder results high local stresses which in turn
566: %dominates in tensile sheet rupture. However in practice breaks are
567: %caused by sticks and pinholes, so the rupture is already initiated and
568: %the breakage is just crack growth. To improve the strength in practice
569: %the disorder is unimportant, possibly the strength of the constituents
570: %is however the key factor \footnote{enhan ma oikeasti ole tuota
571: %mielta}.
572:
573: \begin{thebibliography}{99}
574: \bibitem{sethrev}
575: J.P. Sethna, K.A. Dahmen, and C.R. Myers,
576: Nature (London) {\bf 410}, 242 (2001).
577: \bibitem{Bourev} Scalefree cracks are discussed
578: in: E.\ Bouchaud, J.\ Phys.\ Cond.\ Mat.\ {\bf 9}, 4319
579: (1997).
580: \bibitem{Ciliberto}
581: A. Guarino, A. Garcimartin, and S. Ciliberto,
582: Eur. Phys. J. B {\bf 6}, 13 (1998);
583: A. Garcimartin {\em et al.},
584: %A. Guarino, L. Bellon, S. Ciliberto,
585: Phys. Rev. Lett {\bf 79}, 3202 (1997).
586: \bibitem{penn}
587: L.C. Krysac and J.D. Maynard,
588: Phys. Rev. Lett. {\bf 81}, 4428 (1998).
589: \bibitem{Petri2}
590: A. Petri {\em et al.},
591: %G. Paparo, A. Vespignani, A. Alippi, M. Costantini,
592: Phys. Rev. Lett {\bf 73}, 3423 (1994).
593: \bibitem{Lockner}
594: D.A. Lockner {\em et al.},
595: %J.D. Byerlee, V. Kuksenko, A. Ponomarev, A. Sidorin,
596: Nature {\bf 350}, 39 (1991).
597: \bibitem{oma} L.I. Salminen, A.I. Tolvanen, and M.J. Alava,
598: Phys. Rev. Lett. {\bf 89}, 185503 (2002).
599: \bibitem{simu}
600: J.B. Rundle, S. Gross, W. Klein, C. Ferguson, D.L. Turcotte,
601: Tectonophysics {\bf 277}, 147 (1997).
602: \bibitem{lopinaa}
603: J. Fineberg and M. Marder, Physics Reports {\bf 313}, 1-108 (1999).
604: \bibitem{line} e.g. T.\ Halpin-Healy and Y.-C.\ Zhang, Phys. Rep. {\bf 254},
605: 215 (1995).
606: \bibitem{Fisher} S. Ramanathan, D. Erta\'s, and D.S. Fisher,
607: Phys. Rev. Lett. {\bf 79}, 873 (1997).
608: \bibitem{maloy}
609: J. Schmittbuhl and K.J. Målo$\!\!\!$/y,
610: Phys. Rev. Lett. {\bf 78}, 3888 (1997);
611: A. Delaplace, J. Schmittbuhl, and K. Målo$\!\!\!$/y, Phys. Rev. E {\bf 60},
612: 1337 (1999).
613: K.J. Målo$\!\!\!$/y, {\em et al.}, Int.
614: J. Fract. {\bf 121}, 9 (2003).
615:
616: \bibitem{zapperi}
617: S. Zapperi, H.J. Herrmann, and S. Roux, Eur. Phys. J. B
618: {\bf 17}, 131 (2000);
619: {\.A}str{\"o}m, M. Alava, and J. Timonen, Phys. Rev. E {\bf 62}, 2878
620: (2000); M.J. Alava and S. Zapperi,
621: Phys. Rev. Lett. {\bf 92}, 049601 (2004).
622: \bibitem{yama}
623: T. Yamauchi, S. Okumura, N. Noguchi, Journal of Pulp and Paper
624: Science {\bf 16}, (1990).
625: \bibitem{Zapperi}P. K. Nukala, S. Simunovic, and S. Zapperi,
626: J. Stat. Mech. Theo. Expt., P08001 (2004).
627: \bibitem{recent}
628: M. Minozzi, G. Caldarelli, L. Pietronero, and
629: S. Zapperi, Eur. Phys. J. B {\bf 36}, 203 (2003).
630: \bibitem{herrmann} F. Tzschichholz and H.J. Herrmann,
631: Phys. Rev. {\bf E51}, 1961 (1995).
632: \bibitem{corte}
633: H. Corte, {\em et al.}, in ``The Role of Fundamental
634: Research in Paper Making,'', Ed. J Brander, Mech. Eng.
635: Publ. Ltd., (London, 1983)), pp 571-584.
636: \bibitem{lys}
637: M.V. Lysak, Eng. Fract. Mech. {\bf 55}
638: 443, (1996).
639: \bibitem{week}
640: J.D. Weeks, D.A. Lockner, and J.D. Byerlee, Bull. Seism. Soc. Am.
641: {\bf 68}, 333-341 (1978).
642: \bibitem{Maas}
643: C. Maes, {\em et al.}, Phys. Rev. B {\bf 57}, (1998).
644: \bibitem{geofyysikoita}
645: J. Weiss, J.R. Grasso, and P. Martin,
646: Proc. 6th Int. Conf. on AE/MS in Geol. Struct. \&
647: Mat., 1996, 583-595, Trans Tech Publications,
648: (Clausthal-Zellerfeld).
649:
650: \end{thebibliography}
651:
652: \begin{figure}%\includegraphics[width=7cm%,draft
653: %]{./R.eps}
654: \caption{Photograph of the peel-in-nip device, roll diameter is 80mm.} \label{figyksi}
655: \end{figure}
656:
657: \begin{figure}
658: %includegraphics[width=7cm%,draft
659: %{./F1B.eps}
660: %caption{Fracture scenario in the peel-in-nip test: side view.}
661: %label{figkaksi}
662: \includegraphics[width=7cm%,draft
663: ]{./dencrack4.eps}
664: \caption{Fracture propagation schematics in the peel-in-nip test.}
665: \label{figkaksib}
666: \end{figure}
667:
668: \begin{figure}\includegraphics[width=7cm%,draft
669: ]{./large-den.5.eps}
670: \caption{Example of the data in peel-in-nip test.
671: Both the AE channels, and the force or reeling moment
672: signals are included. The former is in arbitrary units,
673: for the latter see text.} \label{figkaksic}
674: \end{figure}
675:
676:
677: \begin{figure}\includegraphics[width=7cm%,draft
678: ]{./den.gr.3.eps}
679: \caption{Event energy distributions for the peel fracture
680: and comparisons.} \label{figkolme}
681: \end{figure}
682:
683:
684: \begin{figure}\includegraphics[width=7cm%,draft
685: ]{./den.omo.3.eps}
686: \caption{Event interarrival time distributions for the
687: same cases as in the previous figure.} \label{fignelja}
688: \end{figure}
689:
690:
691: %\begin{figure}\includegraphics[width=7cm%,draft
692: %]{./final-leap-2.eps}
693: %\caption{Left subfigure interarrival time distribution in peel-in-nip
694: %test for various strain rates. Right subfigure the same distribtuions
695: %transformed to leap length distributions.} \label{figkuusi}
696: %\end{figure}
697:
698: \begin{figure}\includegraphics[width=7cm%,draft
699: ]{./fig6.eps}
700: \caption{Average energy (arbitrary units)
701: in a window of twenty consecutive
702: events, vs. the average waiting time in the same, for the
703: peel-in-nip test.} \label{figseit}
704: \end{figure}
705:
706: \begin{figure}\includegraphics[width=7cm%,draft
707: ]{./akf.eps}
708: \caption{Autocorrelations of the AE signal for two strain
709: rates. As a comparison, a randomize signal is presented for
710: both cases (see text).} \label{acf}
711: \end{figure}
712:
713:
714: \end{document}
715: