cond-mat0402618/orpe.tex
1: % ****** Start of file apssamp.tex ******
2: %
3: %   This file is part of the APS files in the REVTeX 4 distribution.
4: %   Version 4.0 of REVTeX, August 2001
5: %
6: %   Copyright (c) 2001 The American Physical Society.
7: %
8: %   See the REVTeX 4 README file for restrictions and more information.
9: %
10: % TeX'ing this file requires that you have AMS-LaTeX 2.0 installed
11: % as well as the rest of the prerequisites for REVTeX 4.0
12: %
13: % See the REVTeX 4 README file
14: % It also requires running BibTeX. The commands are as follows:
15: %
16: %  1)  latex apssamp.tex
17: %  2)  bibtex apssamp
18: %  3)  latex apssamp.tex
19: %  4)  latex apssamp.tex
20: %
21: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
22: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
23: 
24: % Some other (several out of many) possibilities
25: %\documentclass[preprint,aps]{revtex4}
26: %\documentclass[preprint,aps,draft]{revtex4}
27: %\documentclass[prb]{revtex4}% Physical Review B
28: 
29: \usepackage{graphicx}% Include figure files
30: \usepackage{dcolumn}% Align table columns on decimal point
31: \usepackage{bm}% bold math
32: \usepackage[pagebackref]{hyperref}
33: 
34: %\nofiles
35: 
36: \begin{document}
37: 
38: \preprint{APS/123-QED}
39: 
40: \title{Solid-fluid transition in a granular shear flow}
41: 
42: \author{Ashish V. Orpe}
43: \author{D. V. Khakhar}%
44: \email{khakhar@iitb.ac.in} \affiliation{Department of Chemical
45:   Engineering, Indian Institute of Technology - Bombay, Powai, Mumbai,
46:   400076, India}
47: 
48: \date{\today}
49: 
50: \begin{abstract}
51:   
52:   The rheology of a granular shear flow is studied in a quasi-2d
53:   rotating cylinder.  Measurements are carried out near the midpoint
54:   along the length of the surface flowing layer where the flow is
55:   steady and non-accelerating. Streakline photography and image
56:   analysis are used to obtain particle velocities and positions.
57:   Different particle sizes and rotational speeds are considered. We
58:   find a sharp transition in the apparent viscosity ($\eta$) variation
59:   with rms velocity ($u$). In the fluid-like region above the depth
60:   corresponding to the transition point (higher rms velocities) there
61:   is a rapid increase in viscosity with decreasing rms velocity. Below
62:   the transition depth we find $\eta \propto u^{-1.5}$ for all the different
63:   cases studied and the material approaches an amorphous solid-like
64:   state deep in the layer. The velocity distribution is Maxwellian
65:   above the transition point and a Poisson velocity distribution is
66:   obtained deep in the layer. The observed transition appears to be
67:   analogous to a glass transition.
68: 
69: \end{abstract}
70: 
71: \pacs{45.70.-n, 45.70.Mg, 83.80.Fg}% PACS, the Physics and Astronomy
72:                              % Classification Scheme.
73: %\keywords{Suggested keywords}%Use showkeys class option if keyword
74:                               %display desired
75: \maketitle
76: 
77: Granular materials are known to exist in solid-like and fluid-like
78: states \citep{jag96}.  Physical understanding of the flow of granular
79: materials has thus developed along two major themes based on the flow
80: regime \citep{jack86}. In the {\it rapid flow \/}| fluid-like |
81: regime, both theory and experimental analysis are generally cast in
82: the framework of the kinetic theory \citep{cam90}. In contrast, the
83: {\it slow flow \/}| solid-like | regime is most commonly described
84: using the tools of soil mechanics and plasticity theory \citep{ned92}
85: and recently by analogy to glasses \citep{danna01a,bar01,kurch00}.  A
86: key difference between solid-like and fluid-like states, which appears
87: to be emerging from recent studies, is the definition of a temperature
88: in the two states. In the fluid-like state the granular temperature is
89: defined as the kinetic energy of velocity fluctuations in analogy with
90: the kinetic theory of gases. In contrast, for dense granular flows
91: (solid-like state), numerical \citep{mak02b} and experimental
92: \citep{danna03b} studies show the validity of the
93: fluctuation-dissipation theorem (the mean-square displacement of a
94: particle is proportional to time for a constant applied force). This
95: allows for the definition of a new temperature \citep{kurch00}. These
96: two approaches have no well-understood region of overlap.  Given the
97: qualitative differences between the fluid and solid states, a question
98: that has been open for some time relates to the criterion for
99: transition between fluid and solid states of granular materials. We
100: focus on this question.
101:   
102: \begin{figure}
103:   \centering{\includegraphics[width=3.45in]{fig1}}
104:   \caption{(a)
105:     Schematic of the rotating cylinder geometry showing the flowing
106:     layer and the co-ordinate system employed. (b) A typical image
107:     showing the portion of the flowing layer at the center of the
108:     cylinder, with streaks of various lengths formed across the layer
109:     for a shutter speed of $1/250$ s.  (c) Magnified image of the
110:     rectangular region marked in (b) showing streaks generated by
111:     three different particles. (d) The same three streaks with the
112:     fitted intensity function.  The optimization yields length ($2l$),
113:     width ($2w$), orientation angle ($\theta$) and position ($x,y$) for
114:     every streak.}
115:   \label{fig:schem}
116: \end{figure}
117: 
118: \begin{figure*}
119:   \centering{\includegraphics[width=6.5in]{fig2}}
120:   \caption{Mean velocity ($\langle v_{x}\rangle$), rms velocity ($u$), number
121:     density ($n$), shear rate ($\dot{\gamma}$) and shear stress ($\tau_{yx}$)
122:     across the flowing layer for $2$ mm steel balls rotated at $3$
123:     rpm. The solid line in (a) shows a linear fit to the mean velocity
124:     profile. The dashed line denotes the transition point in all the
125:     figures.}
126:   \label{fig:basedata}
127: \end{figure*}
128: 
129: There are few studies which focus on the transition between the solid
130: and fluid states. \citet{met02} studied solid to fluid transition in a
131: horizontally vibrated container of beads. They observed hysteresis in
132: the transition which was well-predicted by a dry friction model in
133: which the friction coefficient varies smoothly between a dynamic and
134: static value. A fluid-solid transition was also observed by
135: \citet{danna01a} for vertically vibrated particles.  They found that
136: the vibrating granular medium first undergoes a sharp transition to a
137: super-cooled liquid and then gradually achieves a solid-like state on
138: reducing the intensity of vibration. This process follows a modified
139: Vogel-Fulcher-Tammann (VFT) model which is typical of fragile glasses.
140: 
141: Coexisting solid and fluid phases have been studied primarily in the
142: context of surface flows; these comprise a layer of fluid like flow on
143: a fixed bed of the same material. Examples of surface flows which have
144: been well-studied include heap flows
145: \citep{bouc94,bout98a,gra99,dou99,kom00,lem00,kha01b} and rotating
146: cylinder flows \citep{raj90,elp98,orp01,kha01c,jain02,bon02,hill03}.
147: Remarkably simple theories describe the coexistence between the solid
148: and fluid. In the simplest versions, the local {\it melting\/} and
149: {\it freezing\/} is determined by the local angle of the solid-fluid
150: interface: If the local angle is greater than a ``neutral'' angle, the
151: solid melts (the heap erodes) so as to reduce the angle and vice versa
152: \citep{bouc94,bout98a}.  Continuum models based on Coulombic friction
153: models for the solid region and simple rheological models for fluid
154: region predict a similar behaviour \citep{dou99,kha01b}.  Experimental
155: studies show that the models based on local angle based
156: melting/freezing give good predictions \citep{kha01b}.
157: 
158: An assumption in the coarse-grained models described above is the
159: existence of an interface between the solid and fluid. However, the
160: recent work 
161: of \citet{kom00} indicates that the surface flow on a heap decays
162: smoothly with depth and generates motion deep within the heap.  The
163: interface between the solid and fluid regions is thus not well-defined
164: when considered at a particle length scale.  Studies of the velocity
165: profile in rotating cylinders flow also confirm this picture of a
166: smooth decay of the velocity into the bed, rather than an abrupt
167: change at an interface \citep{jain02,bon02}.
168: 
169: The objective of the present work is to gain an insight into the
170: fluid-solid transition in granular flows focussing on a system where
171: both fluid-like and solid-like regions coexist. We find a well-defined
172: transition point in the system which demarcates two distinct flow
173: regions. The behaviour in each region is characterized and results
174: indicate that the transition has some characteristics of a glass
175: transition.
176: 
177: Experiments are carried out in quasi-2d aluminium cylinders (length
178: $1$ or $2$ cm) of radius $16$ cm (Fig.~\ref{fig:schem}a). The end
179: walls are made of glass and a computer controlled stepper motor with a
180: sufficiently small step is used to rotate the cylinders.
181: Monodisperse, spherical, shiny stainless steel balls of three
182: different sizes with diameters ($d$) $1$, $2$, and $3$ mm are used in
183: the experiments.  Cylinder rotation in the rolling flow regime
184: (rotational speed, $\omega = 2-9$ rpm) produces a thin flowing surface
185: layer and measurements are made near the center of the cylinder where
186: the layer thickness is maximum and the flow is non-accelerating. The
187: particles are heavy enough and conductive so that charge effects are
188: negligible.  The experiments are carried out with $50 \%$ of the
189: cylinder filled with particles.
190: 
191: The motion of the particles is captured by taking high resolution
192: images using a digital camera (Nikon Coolpix 5000) in the presence of
193: an incident beam of light. The size of the recorded region is $2560 \times
194: 1920$ pixels, with one pixel corresponding to $0.016-0.03$ mm
195: depending on the distance of the camera to the cylinder. The point
196: source of light is directed near parallel to the end wall of the
197: cylinder so as to illuminate only the front layer of the flowing
198: particles.  Each moving particle generates a streak of definite length
199: depending on its speed and the shutter speed of the camera. Images are
200: taken for a range of camera shutter speeds ($1/15-1/2000$ s) so as to
201: account for the varying velocity across the flowing layer. This gives
202: streaks, which are adequately long for analysis but not so long as to
203: overlap with other streaks, in each the part of the flowing layer
204: (Fig.~\ref{fig:schem}b). Two hundred images are taken for each shutter
205: speed with an overall of two thousand images combined over different
206: shutter speeds. Due to the time interval between photographs, each
207: experiment typically takes 500 cylinder revolutions.
208: 
209: A parameterized intensity function corresponding to a stretched
210: Gaussian function is then fitted to the intensity values of the streak
211: pixels (and an immediate neighbourhood) in the image by means of a
212: standard non-linear optimization technique. The fitting yields the
213: length ($2l$), width ($2w$), orientation angle ($\theta$) and the position
214: ($x,y$) for the streak (Fig.~\ref{fig:schem}d). The analysis
215: technique was calibrated by 
216: carrying out experiments for a single particle glued to the cylinder
217: end plate. The error in velocity
218: measurements was found to be less than than $3\%$.  For analysis, the
219: flowing layer region is divided into bins of width equal to the
220: particle diameter and length $20$ mm parallel to the flowing layer.
221: The components of the mean and the root mean square (rms) velocities
222: for each bin are calculated by averaging over all streaks in a bin.
223: 
224: Fig.~\ref{fig:basedata} gives the variation of system variables with
225: depth ($y$) in the flowing layer for $2$ mm particles. The mean
226: velocity ($\langle v_{x}\rangle$) profile is smooth and shows 3 regions; a
227: near-linear middle region, an exponentially decaying region at the
228: bottom and a flattened region near the top. Similar profiles have been
229: reported in several previous studies
230: \citep{boc01,longo02,bon02,jain02,mueth03,hill03,tab04,mueth00}. The
231: flattened upper region which is the very low density region
232: corresponding to saltating particles is not seen in some studies. The
233: magnitudes of the rms velocities in the two directions are different:
234: the component in the flow direction ($x$-component) is about 10\%
235: higher than that in direction perpendicular to the flow
236: ($y$-component).  The profile of the total magnitude of the
237: fluctuating velocities ($u$) is shown in Fig.~\ref{fig:basedata}b.
238: The rms velocity profile shows two distinct regions: a relatively slow
239: variation near the free surface followed by a sharper decrease deeper
240: in the bed. Fig.~\ref{fig:basedata}d shows the shear rate variation in
241: the layer, obtained by numerical differentiation of the data in
242: Fig.~\ref{fig:basedata}a.  The shear rate increases to a maximum value
243: (corresponding to the inflection point in the velocity field) and then
244: decreases. We note that an oscillating shear rate profile is obtained
245: if a smaller bin size is used as reported previously
246: \citep{mueth00,mueth03,hill03}. 
247: The transition point in the fluctuation velocity profile
248: coincides with the maximum in the shear rate. The areal number density
249: is almost constant throughout the flowing layer with a rapid decrease
250: near the free surface as shown in Fig.~\ref{fig:basedata}c. 
251: 
252: \begin{figure}
253:   \centering{\includegraphics[width=3.0in]{fig3}}
254:   \caption{Apparent viscosity ($\eta$) variation with rms velocity ($u$)
255:     for all the cases studied. Filled symbols represent the points at
256:     and above the transition velocity ($u_{c}$). (a) The solid line at
257:     the top represents a linear fit (slope $\simeq{1.5}$) to the data below
258:     the transition point for all the cases studied. (b) The data is
259:     scaled using particle mass ($m$), particle diameter ($d$) and
260:     average shear rate obtained by a linear fit to each corresponding
261:     velocity profile.}
262:   \label{fig:eta_u}
263: \end{figure}
264: 
265: \begin{figure}
266:   \centering{\includegraphics[width=3.0in]{fig4}}
267:   \caption{Distributions of velocity in $y$ direction at
268:     various locations across the layer for $2$ mm steel balls rotated
269:     at $3$ rpm. The dashed line with filled circles represents the
270:     transition point, the solid lines represent the region above the
271:     transition point and all the dotted lines represent the region
272:     below the transition point. The thick black line represents a
273:     fitted parabola to the curves at and above the transition point.}
274:   \label{fig:vydist_layer}
275: \end{figure}
276: For
277: non-accelerating flows a force balance yields the shear stress as
278: $d\tau_{xy}/dy = \rho g \sin\beta \approx \rho_{b} (n/n_{b}) g \sin\beta$ where $g$ is
279: acceleration due to gravity, $\beta$ is the angle of repose, $n$ is the
280: number density in the flowing layer, $n_{b}$ is
281: the number density in the rotating packed bed and $\rho_{b}$ is the bulk
282: density of the rotating packed bed. We neglect the contribution of wall
283: friction in the estimate. This could become significant deeper in the
284: bed \citep{tab04}. However, based on the method of \citet{tab04} we
285: find that the contribution is about $10\%$ of the total stress in the
286: cases studied and does not qualitatively affect the results. Upon
287: integration we obtain $\tau_{xy} = (\rho_{b}/n_{b}) g \sin\beta \int_{y}^{0} n dy$
288: assuming $\tau_{xy} = 0$ at the free surface. The shear stress shows a
289: near linear increase with depth (Fig.~\ref{fig:basedata}d). The
290: results of Fig.~\ref{fig:basedata} indicate that there is a
291: qualitative change in the rheology at the transition point ($y_{c}$).
292: Above $y_{c}$, the shear stress increases with shear rate which is
293: typical of fluids. However, below $y_{c}$ the shear stress increases
294: while the shear rate decreases. This implies that the viscosity
295: increases sharply with depth below $y_{c}$, even though the number
296: density is nearly constant.
297: 
298: Fig.~\ref{fig:eta_u}a shows the variation of the apparent viscosity
299: ($\eta = \tau_{xy}/\dot{\gamma}$) with the rms velocity ($u$), considering only
300: the region of constant density. The data points at and above the
301: transition point ($y\geq y_{c}$) are plotted as filled symbols. There is
302: a sharp transition in this case as well and the transition occurs at
303: the same value of $u$ as in Fig.~\ref{fig:basedata}b.  In the region
304: near the free surface there is a rapid increase in viscosity with
305: decreasing rms velocity whereas in the region approaching the fixed
306: bed there is a much slower power law increase with an exponent
307: $n\sim-1.5$ for all the cases studied. The data scales with
308: $\dot{\gamma}_{avg}$ and $d$ to fall on a single curve
309: (Fig.~\ref{fig:eta_u}b), and the transition
310: occurs at $u_c\sim 1.2\dot{\gamma}_{avg}d$ for all cases. Here $\dot{\gamma}_{avg}$
311: is the shear rate obtained by fitting a straight line to the linear
312: portion of the velocity profile (Fig.~\ref{fig:basedata}a).
313: 
314: Fig.~\ref{fig:vydist_layer} shows the distributions of the $y$
315: direction velocities at different locations in the layer. The $v_y$
316: distribution is Gaussian for all points above the transition point
317: ($y\geq y_{c}$) and it gradually evolves
318: to a Poisson distribution as we go deeper into the bed. A Poisson
319: velocity distribution was also found by \citet{mueth03} for Couette
320: flow of a dense granular material. The $v_x$ distribution (not shown)
321: is Gaussian above the transition point. However, below the transition
322: point the behaviour is complex and bimodal distributions are obtained.
323: 
324: The results presented show a sharp transition between two flow
325: regimes. In the upper region near the free surface, the behaviour is
326: fluid-like and the velocity distributions are Maxwellian. Below the
327: transition point the material appears to be an amorphous soft solid,
328: increasing in strength with depth in the layer. The transition to this
329: solid-like regime occurs at a relatively large rms velocity ($\approx 0.1$
330: m/s). We conjecture that the sharp transition occurs because of the
331: formation of a percolated network of particles in extended contact
332: with each other. This is in contrast to the fluid-like regime where
333: the particles interact through collisions. The contact network coexists
334: with fluid-like domains and the fraction of particles which are part
335: of the network increase with depth. The transition appears to be
336: analogous to a glass transition. New rheological models may be
337: required for granular flows that span these two regimes.
338: 
339: \bibliography{/home/ash/rpcs/thesis/ref}
340: 
341: \end{document}
342: