cond-mat0208118/tap.tex
1: %\documentstyle[pre,aps,twocolumn,epsf]{revtex} % obsolete revtex style
2: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: \usepackage{graphicx}% Include figure files                %use with REVTex4
4: \usepackage{dcolumn}% Align table columns on decimal point %use with REVTex4
5: \usepackage{bm}% bold math                                 %use with REVTex4
6: 
7: %\documentclass{article} % plain TeX command
8: 
9: \begin{document}
10: 
11: \title{\bf{Coarsening Dynamics of Granular Heaplets in Tapped Granular Layers}}
12: 
13: \author{Y.~K.~GOH}
14: \author{R.~L.~JACOBS}
15: %\address{Department of Mathematics, Imperial College\\
16: %180 Queen's Gate, LONDON SW7 2AZ, U.K.}
17: % Command \address is used for revtex style format
18: \affiliation{Department of Mathematics, Imperial College\\
19: 180 Queen's Gate, LONDON SW7 2AZ, U.K.}
20: %Command \affiliation is used only for REVTex4 style format
21: \date{\today}
22: 
23: %\maketitle %abstract comes after \maketitle in revtex style
24: 
25: \begin{abstract}
26: A semi-continuum model is introduced to study 
27:  the dynamics of the formation of granular heaplets 
28:  in tapped granular layers.
29: By taking into account the energy dissipation of collisions  and
30:  screening effects due to avalanches, 
31:  this model is able to reproduce qualitatively the pattern of 
32:  these heaplets.
33: Our simulations show that the granular heaplets are characterised
34:  by an effective surface tension which depends on the 
35:  magnitude of the tapping intensity.
36: Also, we observe that there is a coarsening effect in that the
37:  average size of the heaplets, $V$ grows as the number of taps $k$ increases.
38: The growth law at intermediate times can be fitted by a scaling function
39:  $V \sim k^z$ but the range of validity of the power law 
40:  is limited by size effects.
41: The growth 
42:  exponent $z$ appears to diverge as the tapping intensity is increased.
43: \end{abstract}
44: %47.54.+r,45.05.+x,47.35.+i \\
45: \pacs{61.43.Gt,45.70.Qj,83.80.Fg} %use in REVTex4
46: 
47: \maketitle %abstract comes before maketitle in REVTex4
48: 
49: %\section{Introduction}\label{intro}
50: % Why you do this research.
51: % back ground review.
52: In recent years, there has been increasing interest in the collective
53:  dynamics of granular materials.
54: The dissipative nature of granular materials gives rise to properties
55:  distinct from those of solids and liquids.
56: Many experimental and theoretical
57:  attempts have been made to seek a fundamental understanding of this 
58:  granular state and especially pattern formation 
59:  in driven granular layers.
60: The experiments include vertically vibrated 
61:  systems\,\cite{Umbanhowar96,bizon98}, 
62:  tapped or blown thin films of powders\,\cite{Duran00,Duran01}, 
63:  and electrostatically driven granular layers\,\cite{Aranson02,Aranson00}.
64: Although many papers have been published to explain the experiments on
65:  tapped granular layers,
66:  most of them deal with either compactification of 
67:  thick granular layers\,\cite{Ben-Naim98,Knight95}, 
68:  or static properties of granular heaplets on a thin granular
69:  layer\,\cite{Duran01}.
70: The dynamical aspect of the heaping process in a tapped thin granular layer
71:  is still not well understood.
72: This paper is concerned with the coarsening dynamics of 
73:  granular heaplets in tapped granular layers from 
74:  a theoretical point of view.
75: 
76: %introduce what are you going to do.
77: This paper is organised as follows.
78: First we introduce a simple model to study this fascinating system
79:  by considering the energy dissipation  and screening effects during
80:  the tapping process.
81: Despite the simplicity, this model is capable 
82:  of capturing the essential phenomenology, 
83:  reproducing various morphologies of the coarsening pattern, 
84:  and showing the way in which heaplets merge 
85:  as observed in experiments\,\cite{Duran01}. 
86: The model is then studied numerically and the results indicate some of
87:  the relationships between tapping intensity and
88:  the effective surface tension of granular layers.
89: Finally analysis of 
90:  the results shows that the average size of the the heaplets $V$ grows 
91:  as a power law with the number of taps $k$, $V\sim k^z$
92:  for limited range of $k$ values.
93: The exponent $z$ appears to diverge as the tapping intensity is increased.
94: 
95: %\section{Model}\label{model}
96: % 2 phases in tapping process
97: {\em Model} -- 
98: Consider a {\em thin} layer of $N \gg 1$ granular particles
99:  spread out over a flat plate.
100: The plate is then tapped repeatedly at a low pace
101:  with constant shock intensity.
102: In one complete tapping cycle, there are two different processes 
103:  each of which requires a different description.
104: In the tapping phase,
105:  the granular layer is perturbed by an external shock and
106:  granular particles acquire kinetic energy to allow 
107:  them explore the phase space and hop on to neighbouring sites.
108: The hopping range of the granular particles depends on
109:  the amount of kinetic energy received by each of 
110:  the individual particles.
111: The amount of energy supplied to the system is controlled by
112:  the dimensionless acceleration $\Gamma=U/g \, \Delta t$,
113:  where $U$ is the velocity of the plate when it is in motion and
114:  $\Delta t$ is the time interval over which it is in motion.
115: Note that $U$ is also the vertical velocity gained by 
116:  an {\em elastic} particle sitting on the plate when it is tapped.
117: Our particles are not however elastic.
118: Therefore, the vertical velocity gain is $\alpha U$,
119:  where $0 \le \alpha \le 1$ is the departure coefficient
120:  analogous to the coefficient of restitution and 
121:  it characterises the degree of dissipation of the system. 
122: After the tap, the particles undergo a ballistic flight and fall back 
123:  again onto the static plate where 
124:  they relax subject to avalanche dynamics before the next tap.
125: In this relaxation phase, the granular particles may stay immobile
126:  or move about depending on the local slope.
127: If the local slope is less than a critical slope then the profile 
128:  remains stationary.
129: Conversely,
130:  if the local slope is greater than the critical slope,
131:  matter moves down the slope collectively as an avalanche, 
132:  until the slope is less than the critical slope.
133: 
134: %Tapping phase
135: %definition of variables, assumptions made in the model
136: Let the area density $n(\mathbf{x},k)$ be the number of
137:  granular particles above a unit area of the plate after $k$ taps.
138: Here $\mathbf{x} = (x,y)$, where $x$ and $y$ are 
139:  orthogonal coordinates parallel to the plate.
140: The average diameter, $D$ of the granular particles is chosen
141:  to be $1$, so that $n( \mathbf{x},k)$ now is equal to the 
142:  local thickness of the granular layer, so long as there is no
143:  compactification throughout the tapping process.
144: Of course this is just an idealisation, 
145:  the packing fraction of the granular material will be 
146:  changed\,\cite{Nowak98} if the granular layer is thick 
147:  or it is intentionally prepared in a low-density state, but 
148:  such cases will not be considered here.
149: 
150: % single particle description
151: We can approximate the acceleration $A_p$ of the plate's movement
152:  due to tapping as a sum of $M$ delta functions
153:  \begin{equation}\label{eq:A_p}
154:    A_p(t) = \sum_{k=1}^{M}U\, \delta(t-(k-1)\tau),
155:  \end{equation}
156:  where the velocity $U$
157:  gives the intensity of the shocks which are assumed uniform
158:  over the whole plate. $\tau$ specifies the time interval between 
159:  taps. 
160: Here $\tau$ must be greater than any time scale in the relaxation
161:  process, this is to ensure that next tap only occurs
162:  after the system is fully relaxed.
163: In the dilute limit, a single inelastic grain sitting on the plate
164:  experiences a net force 
165:  \begin{equation}\label{eq:F_z}
166:   F_z(t) = \alpha \left( \sum_{k=1}^{M}mU\, \delta(t-(k-1)\tau)\right) - mg,
167:  \end{equation}
168:  where $\alpha$ is the departure coefficient mentioned earlier.
169: During a small interval 
170:  $[t_k - \Delta t/2, t_k + \Delta t/2]$, 
171:  where $t_k$ is the short hand for $t_k = (k-1)\tau$, the time
172:  of $k$-th tap, and the maximum velocity transferred to the particle is
173:  \begin{equation}\label{eq:v_z}
174:    \Delta v_z = \int_{t_k - \frac{\Delta t}{2}}^{t_k + \frac{\Delta t}{2}}
175:        \frac{F_z(t)}{m} \,  \mathrm{d}t = 
176:    \alpha U - g \, \Delta t.
177:  \end{equation}
178: In order for the particle to hop, $\Delta v_z$ must be positive,
179:  or $\alpha U > v_0 \equiv g\, \Delta t$
180:  so that external acceleration overcomes the gravitational force.
181: In what follows this is always the case.
182: 
183: % Continuum limit modification
184: In the continuum limit one has to make two crucial modifications.
185: $\alpha$ must be replaced by an {\em effective} departure coefficient
186:  of the granular bulk. 
187: $\alpha$ is expected to be a monotonic decreasing function of 
188:  $n(\mathbf{x},k)$.
189: This is because when the number density is high,
190:  inter-grain collisions occur more often and more energy is dissipated,
191:  so that grains depart with a smaller departure velocity.
192: While the precise form of $\alpha(n)$ is unknown, in this 
193:  paper it is taken to be 
194:  \begin{equation}\label{eq:alpha}
195:    \alpha(n) = \left\{
196:     \begin{array}{ll}
197:        \frac{\alpha_0 \,\bar{n}}{n(\mathbf{x},k)},
198:      & n > 1 
199:      \\
200:        \alpha_0 \bar{n},
201:      & n \le 1,
202:      \\ 
203:     \end{array}
204:     \right.
205:  \end{equation}
206:  for reasons of simplicity and to make comparison with Duran's 
207:  model\,\cite{Duran01}.
208: Here, $\bar{n}$ is the average density of the system and
209:  $\alpha_0$ is the effective departure coefficient for
210:  average density $\bar{n}$. 
211: Also, the second term in Eq.(\ref{eq:F_z}) needs to be changed.
212: In the continuum limit, according to Duran\,\cite{Duran00}
213:  there is an effect dependent on the position of a grain in the heap.
214: If the grain is not at the top of the heap it supports a fraction of the weight
215:  of the grains above it, 
216:  hence its effective mass is increased.
217: As a result, in order for a granular particle sitting on the inclined 
218:  side of the granular layer to hop,
219:  it requires a larger velocity kick than a particle sitting
220:  on the flat region of the layer.
221: This screening effect can be incorporated into Eq.(\ref{eq:F_z}) by
222:  replacing $m$ in the second term with an effective mass
223:  \begin{equation}\label{eq:mass}
224:    m^* = \left((n_T - n) p \,\sin \theta_c /D \right) m.
225:  \end{equation}
226: Here $D$ is the average grain diameter and it is set to 1 henceforth.
227: $n_T$ is the altitude of the nearest peak in the corrugated granular layer.
228: $p$ is a parameter of unknown value which is set equal to 5 in 
229:  reference~\cite{Duran00}
230:  and this value appears to give a match to experimental results.
231: $\theta_c$ is the angle of repose 
232:  and is close to the value of $\pi/6$ \,\cite{Duran00}.
233: After these modifications, Eq.(\ref{eq:v_z}) now can be written as
234:  \begin{equation}\label{eq:dv}
235:  \frac{\Delta v_z}{v_0} = \left\{
236:   \begin{array}{ll}
237:      \frac{\mathcal{A}}{n} - (n_T - n)p\,\sin\theta_c
238:    & n > 1 
239:    \\
240:      \mathcal{A}- (n_T - n)p\,\sin\theta_c
241:    & n \le 1,
242:    \\
243:   \end{array}
244:   \right.
245:  \end{equation}
246: where $\mathcal{A} = \frac{\alpha_0 \bar{n} U}{v_0}$ is 
247:  the tapping intensity.
248: 
249: % Comparison with Duran's Model
250: There are close similarities between our model and Duran's~\cite{Duran01}.
251: In both models, the resulting pattern formation is due to the competition 
252:  between the upward hopping motion of particles and the downward screening 
253:  effect of avalanches.
254: However, the mechanism causing upward motion is different in the two models. 
255: Duran~\cite{Duran01} and Shinbrot~\cite{Shinbrot98} 
256:  conjectured that the hopping of granular particles is due to the
257:  upcoming air-flux through the porous bed of the granular layer.
258: The velocity of the air-flux at height $n$ can be approximated by
259:  Darcy's law $v_{air}=Kp/n$, 
260:  where $K$ is the permeability of the granular bed, 
261:  $p$ is the pressure different across the granular bed.
262: The velocity of the air-flux $v_{air}$
263:  must be greater than the free fall velocity
264:  of a granular particle $v_{f}$ in order to eject this particle.
265: However, in our model we adopt a simpler picture where
266:  kinetic energy is transferred to the granular  particles 
267:  via direct collisions between plate and particle and
268:  between particle and particle.
269: Energy is dissipated via the effective departure coefficient.
270: The form of the effective departure coefficient is chosen as
271:  in Eq.(\ref{eq:alpha}) so that the velocity transfer from plate to 
272:  the particle
273:  has the same $1/n$ dependency ($n \ge 1$) as $v_{air}$ in Darcy's law.
274: Obviously, we have the freedom to choose other functional forms of $\alpha(n)$,
275:  the simple choice here is just to achieve comparison with Duran's model.
276: As observed from simulation, however, it turned out that the exact form of 
277:  $\alpha(n)$ is unimportant so long as it is a monotonic decreasing function.
278: 
279: % Hopping process
280: Now it remains to specify the hopping process in the tapping phase.
281: When the granular layer is tapped, at an unstable site where $\Delta v_z > 0$,
282:  particles can hop to 
283:  any neighbouring site within a circle defined by a maximum
284:  horizontal hopping range, $R_{max}({\mathbf \Delta v})$.
285: Here we assume on the unstable site,
286:  all the granular particles are re-distributed.
287: Each grain hops with equal probabilities in any direction and
288:  with a random hopping range, $R<R_{max}$.
289: We can estimate the maximum hopping range
290:  $R_{max}({\mathbf \Delta v})$ by the following 
291:  simple arguments.
292: The maximum velocity received by a particle during one tap is 
293:  $\Delta v_z$ vertical to the plate.
294: In this case $R_{max}({\mathbf \Delta v})$ is zero
295:  for the motion is strictly vertical.
296: It is unlikely that particles will always hop in the 
297:  vertical direction, an inter-particle collision can change the
298:  hopping direction to any angle.
299: Assume that particle is ejected with a 
300:  speed $\Delta v_z \cos\phi$ with $\phi$ denoting 
301:  the angle from the vertical axis of the plate.
302: Then, the horizontal hopping range is
303:  \begin{equation}\label{eq:R}
304:    R = \Delta v_z \cos\phi \sin\phi \,t'
305:      = \frac{2 \cos^3\!\phi \sin\phi}{g} (\Delta v_z)^2,
306:  \end{equation}
307: where $t'$ is the flight time of the particle in vacuum.
308: $R$ is maximum when $\phi = \pi/6$, so that the estimated value for $R_{max}$
309:  is given by
310:  \begin{equation}\label{eq:R_max}
311:    R_{max}(\Delta v_z) = \frac{3\sqrt{3}}{8g}(\Delta v_z)^2
312:                        = \xi (\frac{\Delta v_z}{v_0})^2
313:  \end{equation}
314:  where 
315:  \begin{equation}\label{eq:xi}
316:    \xi = \frac{3\sqrt{3}}{8}v_0\,\Delta t.
317:  \end{equation}
318: The estimated $R_{max}$ is very crude and does not take into
319:  account the details of the air-particle and particle-particle interactions
320:  except phenomenologically.
321: However, one can consider $\xi$ in $R_{max}$ as an adjustable
322:  parameter, which varies from one experiment to another experiment
323:  depending on the roughness of the material used in experiment
324:  and humidity of the surrounding environment.
325: 
326: %Relaxation phase
327: There are several ways to describe the avalanche process 
328:  during the relaxation phase.
329: Here we use a slope dependent diffusion equation.
330: We start from the equation of continuity
331:  $\partial_t{n} = - \mathbf{ \nabla \cdot J}$ with the constitutive
332:  current $\mathbf{J}$ given by
333:  \begin{equation}\label{eq:J}
334:    \mathbf{J} = \left\{
335:    \begin{array}{ll}
336:      -\eta(\beta |\nabla n|^2 - 1)\nabla n, 
337:      &
338:      |\nabla n| > \frac{1}{\sqrt{\beta}} 
339:      \\
340:      0, &
341:      |\nabla n| \leq \frac{1}{\sqrt{\beta}} ,
342:    \end{array}
343:    \right.
344:  \end{equation}
345:  where $1/\sqrt{\beta} = \tan\theta_c$ is the critical slope
346:  and $\eta$ sets the diffusion rate.
347: The current is chosen in such a way that when the gradient is 
348:  greater than the critical slope, mass current flows down-hill
349:  to smooth out density fluctuations, 
350:  but no mass is transferred if the gradient is less than the critical slope.
351: 
352: %\section{Simulation}
353: % Method, setup
354: {\em Simulation} -- 
355: The model is studied on a $N \times N$ square lattice with a periodic
356:  boundary condition.
357: Initially, the granular layer is prepared by assigning a random height
358:  between $[0,5]$ to each site and letting the granular layer relax.
359: At the beginning, the fluctuation in $n(\mathbf{x},0)$ is considerably 
360:  smaller then the height of the heaplets formed later.
361: On each tapping cycle, the velocity of particles at each site is calculated 
362:  and the density number is updated according to 
363:  rules defined by Eq.(\ref{eq:dv}) and Eq.(\ref{eq:R_max}).
364: After which the equation of continuity and 
365:  the corresponding constitutive current 
366:  equation Eq.(\ref{eq:J}) are solved and iterated numerically until 
367:  there is no more mass is transferred at each site.
368: 
369: % Results: Pattern description
370: The following density plots show typical results from the simulation.
371: Fig.\ref{fig:densityPlot} shows the density plots 
372:  of $n(\mathbf{x},k)$ at different values of $k$, the number of taps,
373:  for two set of parameters.
374: The top panel shows extended patterns of ridges and
375:  corresponds to a smaller tapping intensity ($\mathcal{A}=6.0$), 
376:  while the lower panel shows localised heaplets and corresponds to 
377:  a greater tapping intensity ($\mathcal{A}=16.8$).
378: Both simulations coarsen when $k$ increases, eventually 
379:  reaching their final stage where the entire system is a single 
380:  granular heap.
381: 
382: One can see from Fig.\ref{fig:densityPlot}
383:  that as the tapping intensity is increased, 
384:  the patterns favour isolated circular heaplets 
385:  which suggests that an effective surface tension can be defined
386:  which increases with the tapping intensity.
387: Since there are no attractive forces between granular particles,
388:  the granular layer cannot have a true surface tension.
389: This ``surface tension'' effect is entirely due to the system trying to
390:  maximise the local energy dissipation by decreasing 
391:  the value of the effective departure coefficient 
392:  so that particles are less mobile.
393: In order to maximise the local energy losses, 
394:  particles have to be as close to each other as possible, 
395:  for this will increase the
396:  number of inelastic collisions between particles.
397: Therefore, with the constraint that the local slope should
398:  not exceeds the critical slope,
399:  the global attractor of the pattern is a single large heap, 
400:  where local energy loss is the largest.
401: Akiyama {\em et al.}\,\cite{Akiyama98},
402:  Cl\'ement\,\cite{clement92} and Duran\,\cite{Duran01}
403:  have discussed the effects of convection on heaping process of 
404:  granular layers.
405: Duran\,\cite{Duran01} in particular suggests that 
406:  the ``surface tension'' is brought about by the convective forces
407:  dragging particles from the surroundings into the granular heaplets. 
408: However, in our simulation it is due to the maximisation  of energy losses, 
409:  since in our model the granular layer does not contain any information
410:  about the internal interactions of the sand heap,
411:  therefore convective forces are not taken into account.
412: 
413: \begin{figure}[!htbp]
414:  \vspace{0.3cm}
415:  \includegraphics[width=8cm]{fig_1.eps}
416:  \caption{\label{fig:densityPlot}
417:   Two simulations of coarsening dynamics of tapped granular layers
418:   corresponding to different tapping intensities.
419:   The simulation images (a)--(d) and (e)--(h) start with
420:   the same initial configuration.
421:   Figures (a)--(d) use a smaller tapping intensity
422:     ($\mathcal{A}=6.0, \xi=0.2207$), 
423:     and Figures (e)--(h) correspond to a stronger tapping intensity 
424:     ($\mathcal{A}=16.8, \xi=0.2207$).
425:   (a) and (e) are taken at $k=5$, 
426:   (b) and (f) at $k=10$,
427:   (c) and (g) at $k=20$,
428:   (d) and (h) at $k=30$.
429:  }
430: \end{figure}
431: 
432: \begin{figure}[!htbp]
433:  \vspace{0.3cm}
434:  \includegraphics[width=8cm]{fig_2.eps}
435:  \caption{\label{fig:merge}
436:    Merging of two granular heaplets. $\mathcal{A}=16.0, \xi=0.2207$.
437:    When two heaplets touch each other, smaller heaplet is sucked into 
438:    the larger one.
439:    (a) $k=15$ two granular heaplets meet.
440:    (b) $k=30$ smaller heaplet is sucked into the larger one.
441:    (c) $k=50$ small heaplet disappears.
442:  }
443: \end{figure}
444: 
445: % Rayleigh-Taylor instabilities & Laplace-Young Pressure
446: Nevertheless, the simulation result does 
447:  show the Laplace-Young pressure effect suggested by reference\,\cite{Duran01}.
448: A typical example is shown in Fig.\ref{fig:merge}.
449: (The parameters here are $\mathcal{A}=16.0$ and $\xi=0.2207$,
450:  but similar results are found over a wide range of 
451:  values of these parameters.)
452: In the center of the figure there are two connected granular heaplets.
453: After several taps the smaller heaplet merges with
454:  the larger heaplet due to matter moving along the connecting neck.
455: 
456: % Dynamics of coarsening
457: We are also interested in the dynamics of the size of the heaplets.
458: The relevant length scale measure $l$ is the average 
459:  thickness of the site $h_i$ weighted by the local volume, $V_i$,
460:  and the average volume of the heaplet is $V=l^3$.
461: Since we know that $h_i=n_i$ and the local volume of the site is just 
462:  $V_i = n_i \sigma$, where $\sigma$ is the area element of the lattice site,
463:  it follows that
464:  \begin{equation}\label{eq:length}
465:    l = \frac{\sum_i \,h_i\, V_i}{\sum_i V_i} 
466:      = \frac{\sum_i n_i^2}{\sum_i n_i}.
467:  \end{equation}
468: 
469: \begin{figure}[!htbp]
470:  \vspace{0.6cm}
471:  \includegraphics[width=8cm]{fig_3.eps}
472:  \caption{\label{fig:LogLog}
473:   Log-log plot for characteristic heaplet size $V$ against number of taps $k$
474:     for varying tapping intensities $\mathcal{A}$.
475:   The solid line corresponds to $\mathcal{A} = 18.0$,
476:     the two dashed lines to $\mathcal{A} = 17.6$ 
477:     the dash-dotted line to $\mathcal{A} = 16.0$,
478:     and the dotted line to $\mathcal{A} = 14.4$.
479:   The parameter used in the simulation is $\xi = 0.2207$.
480:   All curves correspond to a lattice size $L_1 \times L_1 = 80 \times 80$
481:     and are averaged over ten different runs except for
482:     the marked dashed curve which corresponds to 
483:     $L_2 \times L_2  = 120 \times 120$ and is averaged over five runs.
484:   Each curve saturated after a long time, 
485:     this can be clearly seen from the $\mathcal{A}=18.0$ solid line.
486:   The marked curve corresponds to the larger lattice size 
487:    shows a wider range of the 
488:    linear region and a larger saturation value of $V$.
489:  }
490: \end{figure}
491: 
492: There is some limited evidence for power law behaviour $V \sim k^z$
493:   in the intermediate region of the log-log plots of $V$ versus $k$ in
494:   Fig.\ref{fig:LogLog}.
495: There are three regions in each plot.
496: The early region is where the granular layer is randomly distributed, 
497:  the number density fluctuation is not large enough to trigger coalescence.
498: The late-time region is where all the heaplets have joined into one big heap,
499:  leaving some remaining individual particles
500:  hopping randomly about and occasionally encountering the large
501:  single heap.
502: The intermediate region is also the fast-growing region where heaplets
503:  grow and merge into each other.
504: The range of the intermediate region grows and remains linear as the system 
505:  size increases as can be seen from the two curves with 
506:  $\mathcal{A}=17.6$.
507: This shows that system size limits the range of the intermediate 
508:  power law region.
509: For very large system sizes we would expect on this basis that
510:  the intermediate power law region would have a greater 
511:  range and remain linear in the log-log plot.
512: %In this limited region we can parametrise the growth law of the average
513: % size of the heaplets $V$ as $V \sim k^z$ and $z$ varies for different 
514: % tapping intensities.
515: %Note that the system size limits the range of this power law region.
516: %For larger systems, the intermediate linear portion of the curve has 
517: % greater range as shown in the marked dashed line in Fig.\ref{fig:LogLog}.
518: One can easily estimate the increase in
519:  the average volume of the heaplets $V$ by the 
520:  following simple argument.
521: As the saturated value of $V$ corresponds to a state where a single heap 
522:  contains approximately all the grains in the system, 
523:  it should be proportional to $\bar{n} L^2$,
524:  here the average number of  grains per site $\bar{n}$ is a conserved quantity.
525: Then, increment in $V$ is given by $\Delta\log_{10}V =2\,\Delta\log_{10} L$.
526: In Fig.\ref{fig:LogLog} 
527:  the dashed line corresponds to
528:  a lattice size $L_1 \times L_1=80 \times 80$ and 
529:  the marked-dashed line to $L_2 \times L_2 = 120 \times 120$,  
530:  and the vertical shift in the graphs is expected to be
531:  $\Delta \log_{10}V = 0.352$
532:  which may be compared to $0.360$ from the simulation.
533: 
534: % Responds of z to parameter 
535: \begin{figure}[!hbp]
536:  \vspace{0.5cm}
537:  \includegraphics[width=7cm]{fig_4.eps}
538:  \caption{\label{fig:slope_A}
539:   Exponent $z$ for varying tapping intensity $\mathcal{A}$.
540:   For small $\mathcal{A}$ ($\mathcal{A}< 14$), $z$ is roughly constant 
541:   ($z \sim 1.0$), 
542:   but appears to diverge for $\mathcal{A} > 14$.
543:   Each value of $z$ is averaged over ten different runs.
544:  }
545: \end{figure}
546: The exponent $z$ in the power law depends on the tapping 
547:  intensity $\mathcal{A}$ as is shown in Fig.\ref{fig:slope_A}
548: %Fig.\ref{fig:slope_A} shows the variation of $z$ for different $\mathcal{A}$.
549: As we can see $z$ is roughly constant, $z\sim 1.0$,
550:   for low tapping intensity. 
551: When the tapping intensity is increased $\mathcal{A}>14$,
552:  $z$ suddenly increases,  and appears to diverge
553:  near $\mathcal{A}=18$.
554: For these large values of the tapping intensity, we observed that the system
555:  appears to be in a flat configuration for some time 
556:  and suddenly a single heap formed within a few taps.
557: For still greater tapping intensities,
558:  the heaplet patterns disappear and the layer remains flat
559:  with a small random variation superimposed.
560: This pattern-disorder transition occurs at a not very precisely defined
561:  critical tapping intensity of $\mathcal{A} \sim 18$. 
562: (This critical intensity appears to depend slightly on the 
563:  initial conditions which is the reason for our statement that 
564:  it is not precisely defined.)
565: Above this critical tapping intensity, no patterns are observed.
566: This is due to the fact that the perturbation is so strong that 
567:  no site is stable, 
568:  and each site constantly undergoes re-distribution on each tapping cycle.
569: Although the power law parametrisation is based on limited 
570:  evidence, it does provide a useful parameter $z$ to distinguish the 
571:  pattern-forming region ($\mathcal{A} < 18$) from the non-pattern-forming
572:  region ($\mathcal{A}>18$).
573: 
574: %Conclusion
575: We have introduced a simple model of the heaping process 
576: in a tapped granular layer.
577: The model is capable of reproducing the essential morphologies of tapped
578:  granular systems.
579: Qualitatively the effective surface tension of the granular heaps 
580:  is closely related to the tapping intensity,
581:  and it is shown that there is no need for convective forces 
582:  for this effective surface tension to exist.
583: The length scale of the system coarsens according to 
584:  the power law $l\sim k^z$ in a limited range.
585: The exponent $z$ appears to diverge as tapping intensity is increased
586:  and provides a useful parameter for distinguishing the pattern-forming 
587:  and the non-pattern-forming regions.
588: 
589: %Acknowledgment
590: We thank Philip Cheung for helpful discussions.
591: 
592: \begin{thebibliography}{99}
593: 
594:   \bibitem{Umbanhowar96}
595:    P.~B.~Umbanhowar, F.~Melo, and H.~L.~Swinney,
596:    Nature(London) {\bf 382},793 (1996);
597: 
598:   \bibitem{bizon98}
599:    C.~Bizon {\em et al.},
600:    Phys. Rev. Lett. {\bf 80}, 57 (1998).
601: 
602:   \bibitem{Duran01}
603:    J.~Duran, 
604:    Phys. Rev. Lett. {\bf 87}, 254301 (2001).
605: 
606:   \bibitem{Duran00}
607:    J.~Duran, 
608:    Phys. Rev. Lett. {\bf 84}, 5126 (2000).
609: 
610:   \bibitem{Aranson02} 
611:    I.~S.~Aranson, B.~Meerson, P.~V.~Sasorov and V.~M.~Vinokur,
612:    Phys. Rev. Lett. {\bf 88}, 204301 (2002);
613: 
614:   \bibitem{Aranson00}
615:    I.~S.~Aranson {\em et al.},
616:    Phys. Rev. Lett. {\bf 84}, 3306 (2000).
617: 
618:   \bibitem{Ben-Naim98}
619:    E.~Ben-Naim, J.~B.~Knight, E.~R.~Nowak, H.~M.~Jaegar and S.~R.~Nagel,
620:    Physica D {\bf 123}, 380 (1998).
621: 
622:   \bibitem{Knight95}
623:    J.~B.~Knight, C.~G.~Fandrich, C.~N.~Lau, H.~M.~Jaegar and S.~R.~Nagel,
624:    Phys. Rev. E {\bf 51}, 3957 (1995).
625: 
626:   \bibitem{Nowak98} 
627:    E.~R.~Nowak, J.~B.~Knight, E.~Ben-Naim, H.~J.~Jaeger  and S.~R.~Nagel, 
628:    Phys. Rev. E {\bf 57}, 1971 (1998). 
629: 
630:   \bibitem{Shinbrot98}
631:    T.~Shinbrot,
632:    Granular Matter {\bf 1}, 145 (1998).
633: 
634:   \bibitem{clement92}
635:    E.~Cl\'ement, J.~Duran and J.~Rajchenbach.
636:    Phys. Rev. Lett. {\bf 69}, 1189 (1992).
637: 
638:   \bibitem{Akiyama98}
639:    T.~Akiyama, K.~M.~Aoki, K.~Yamamoto and T.~Yoshikawa,
640:    Granular Matter {\bf 1}, 15 (1998)
641: 
642: \end{thebibliography}
643: 
644: \end{document}
645: