Você está na página 1de 9

(This is a sample cover image for this issue. The actual cover is not yet available at this time.

This article appeared in a journal published by Elsevier. The attached


copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy

Computers & Geosciences 49 (2012) 270–277

Contents lists available at SciVerse ScienceDirect

Computers & Geosciences


journal homepage: www.elsevier.com/locate/cageo

Numerical simulation of inclination shallowing by rolling and slipping


of spherical particles
J. Jezek a,n, S. Gilder b, D. Bilardello b
a
Institute of Applied Mathematics and Information Technologies, Faculty of Science, Charles University in Prague, Albertov 6, 128 43 Praha 2, Czech Republic
b
Department of Earth and Environmental Sciences, Ludwig Maximilian University of Munich, Theresienstrasse 41, 80333 Munich, Germany

a r t i c l e i n f o a b s t r a c t

Article history: We report algorithms for two- and three-dimensional numerical simulations of settling spherical
Received 4 April 2012 magnetic particles with prescribed size-distributions. Particles roll, or roll and slip, on the substrate,
Received in revised form which causes their magnetic moments to rotate. These models are applied to the problem of inclination
14 June 2012
shallowing, which is repeatedly encountered in paleomagnetic studies of sedimentary rocks, where the
Accepted 15 June 2012
recorded inclination is less than the expected field inclination. Simulations of equal-sized assemblages
Available online 21 June 2012
of magnetic spheres yield shallowing factors of 0.6, similar to that found in nature and in laboratory
Keywords: redeposition experiments. Comparable results are obtained when varying the size distributions of the
Random settling spheres. Inclination shallowing is more pronounced when the smaller particles are magnetic and the
Roll-slip
larger ones are non-magnetic. Our study shows that rolling and/or slip (translation) of spherical
Spherical particles
particles can significantly contribute to inclination shallowing.
Sedimentary paleomagnetism
Inclination shallowing & 2012 Elsevier Ltd. All rights reserved.

1. Introduction interaction, rolling and possibly experiencing shear from bottom


currents. Particle shape (spherical, prolate, oblate) must also be
Sedimentary rocks are commonly sampled for paleomagnetic considered (Arason and Levi, 1990; Jezek and Gilder, 2006). Within
studies because they have well defined paleohorizontal surfaces and the sediment column, bioturbation, dewatering, diagenesis and
their magnetic signals accumulate over time scales long enough to compaction can modify the magnetization, which is known as a
average out secular variation of the geomagnetic field. On the other post-depositional remanent magnetization (pDRM) (see Tarling and
hand, deposition and lithification processes often result in a Turner, 1999 and references therein).
recorded inclination that is shallower than the ambient field Laboratory redeposition experiments yield highly hetero-
inclination. This effect, known as inclination shallowing, was recog- geneous results depending on the experimental conditions (mate-
nized early in paleomagnetism. Theoretical treatment of inclination rial used, fluid salinity, etc.). The only experiments carried out on
shallowing first considered the role of spherical particles as they solely spherical beads show that inclination shallowing varies
dropped through the water column, hit the bottom, and then rolled with applied field with flattening factors (f, called also the
into depressions (King, 1955; Griffiths et al., 1960). inclination shallowing factor), which is the ratio of tangents of
Acquisition of remanent magnetization in sediments, called a the recorded inclination to the applied inclination, ranging from
depositional remanent magnetization (DRM), is a complex pro- 0.5 at 25 mT and 0.7 at 100 mT, with an average of 0.6 (Bilardello
cess. Falling magnetic particles are subject to balanced inertial, et al., 2011 and 2012). In redeposition experiments of synthetic
viscous and magnetic torques, and can attain alignment with the material composed of magnetic carriers mixed with clay or quartz,
ambient field within seconds (Nagata 1961; Collinson, 1965). van Vreumingen (1993) found that f is about 0.6 (maghemite-
Nevertheless, other factors come into play such as mutual particle dominated) or 0.85 (magnetite-dominated) for suspensions in
interactions, Brownian motion, van der Waals forces, flocculation, low salinity ( o5 g/l) fluids. Essentially no flattening (f¼1.0) was
etc. (Verosub et al., 1979; Blow and Hamilton, 1978; Shcherbakov found when salinities reached seawater levels (35 g/l). King
and Shcherbakova, 1983; Tauxe and Kent, 1984; Anson and Kodama, (1955) determined an average f factor of 0.4 from experiments
1987; Tauxe et al., 2006; Scherbakov and Sycheva, 2010). Eventually conducted on natural silts over a range of applied field inclina-
the particles encounter the substrate, leading to mechanical tions. Tauxe and Kent (1984) redeposited natural sediment in
deionized water. They found the inclination shallowing to be
independent of the applied field with an average f value of 0.55
n
Corresponding author. Tel.: þ420 2 2195 1580; fax: þ420 2 2195 1429. for all applied field inclinations. These are but a few examples of
E-mail address: jezek@natur.cuni.cz (J. Jezek). the myriad of available experimental data.

0098-3004/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.cageo.2012.06.013
Author's personal copy

J. Jezek et al. / Computers & Geosciences 49 (2012) 270–277 271

In this contribution we treat the settling and rolling of solely rotated to produce the final magnetic remanence in a sediment.
spherical particles, which serves as an end-member of the compli- Simulation of random settling of spheres was already studied to
cated process of magnetic field recording in sediments. The model of model packing and porosity of sediments (e.g., Torry et al., 1973;
Griffiths et al. (1960), which is one of basic models of inclination Weltje and Alberts, 2011 and citations therein). Those works did
shallowing, assumes that spherical particles roll around horizontal not study the rotation of (magnetic) particles during deposition. To
axes once they hit the bottom. The azimuths of the rotation axes are our knowledge, no study has treated the role of slip in magnetic
equally distributed, i.e., the particles rotate with equal probability in recording.
any direction. When averaging an assemblage of identical spherical
particles that have an initial magnetic vector [xF,yF,zF] oriented
parallel to the magnetic field and roll through the same angle of
2. A two dimensional model of randomly settling and
rotation (a) along all azimuths, the average vector, i.e., the observed
rolling spheres
magnetization, is:
x ¼ xF ð1 þ cos aÞ=2, y ¼ yF ð1þ cos aÞ=2; z ¼ zF cos a ð1Þ Finding a representative value of a is difficult because settled
spheres become incorporated within the substrate onto which
The inclination shallowing factor (f) is defined as:
new spheres are deposited. Depending on the sedimentation
tanIO 2cos a process, settling particles continually create surfaces with time-
f¼ ¼ ð2Þ
tanIF 1þ cos a varying topographies. We will first consider the settling of equal-
sized, magnetic remanence-carrying spheres as shown in Fig. 2,
where IF is the field inclination and IO is the observed inclination,
where motion is allowed only in the x  z plane. A new sphere is
and
  generated randomly in the x-direction from above, and then falls
2cos a until it hits a stationary sphere at the bottom. Here, it must be
IO ¼ arctan tanIF ¼ arctan ðf tanIF Þ ð3Þ
1 þ cos a decided if, or how, the sphere will roll. The sphere rolls in small
The relative magnetic moment (Mr) is defined as: increments until it reaches a hollow where it stops. The algorithm
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi uses an auxiliary envelope above the stationary spheres whose
MO cosIF points are at a distance R¼ rþr from the centers of the uppermost
Mr ¼ ¼ ð1 þcos aÞ2 þ ð2cos atanIF Þ2
MF 2 stationary spheres (dashed line in Fig. 2, r is the particle radius).
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cosIF The envelope is composed of segments of circles having radius R.
¼ ð1þ qÞ2 þð2qtanIF Þ2 ð4Þ
2 Rolling of a sphere is equivalent to moving its center on the
where, MO is the observed moment and MF is the moment in the envelope.
absence of inclination shallowing, and q¼f/(2 f). Mr is a unitless For each increment, the magnetization vector of the rolling
value between 0 and 1. sphere rotates around a horizontal axis perpendicular to the x  z
A limitation of this model, which has persisted since its plane by an angle Da ¼2 ds/R, where ds is the distance by which
original formulation (Griffiths et al., 1960), is not knowing a the center of the rolling sphere is shifted on the auxiliary
representative value for the angle a. This lack of information is envelope. We consider the rotated vector as one of many vectors
accented by the fact that, according to the above equations, that, according to the assumption of Griffiths et al. (1960), should
inclination shallowing and relative moment change slowly with rotate with equal probability about horizontal axes only. We thus
a when a o50o (Fig. 1). Other limitations of the model are that the compute the final horizontal and vertical component of the vector
spheres are assumed to be identical in size and that no translation from equation (1). During the simulation, we compute the
(slip) of the particles occurs. Below we present algorithms over- accumulated magnetization and the parameters f, a, Io, and Mr
coming these obstacles and give the main results of simulations. after omitting the contribution from the first (initial, bottommost)
Our models combine random settling with rolling and slipping of row of spheres. Although the magnetic moments of particles may
spherical magnetic particles, by which magnetic vectors are be somewhat misaligned when the particles encounter the
bottom, we typically prescribe perfect orientation of the moments
parallel to the magnetic field direction because we are interested
in the net effect of rolling. Note also that the falling particle
alignment parallel to the magnetic field is relatively very fast
compared to the settling time (Nagata, 1961; Scherbakov and
Scherbakova, 1983; Jezek and Gilder, 2006). A Matlab implemen-
tation of the algorithm is presented in Appendix A.

Fig. 1. Inclination shallowing factor (f, thick line), relative change of inclination
(thin lines) and magnetic moment (dashed lines) as function of angle of rotation Fig. 2. A sample run of the 2D model of random sedimentation with rolling. Bars
(a) according to the model of Griffiths et al. (1960). Individual lines for inclination indicate magnetic moments. Dashed line is the auxiliary envelope at a distance
and moment are for field inclination (IF) 10o, 20o,y, 80o. rþ r from the centers of the uppermost stationary spheres.
Author's personal copy

272 J. Jezek et al. / Computers & Geosciences 49 (2012) 270–277

Fig. 3 shows an example that demonstrates how f, a, Io, and Mr 3. Regular periodic schemes
vary when each new sphere stops rolling; the parameters become
nearly constant after about 10 layers of spheres become fully Before progressing to a full 3D model, we first consider the
settled. The total width (x direction) of the domain plays a role regular periodic schemes composed of equal spheres lying along a
and must be chosen sufficiently large, at least several tens of plane with their centers arranged in a triangular or square net
spheres, to obtain stable results. The algorithm and Eqs. (2) to (4) (Fig. 4). When a sphere falls on the structure, it rolls toward the
dictate that f is constant for any orientation of the magnetic field. topographically lowest possible position, or ‘‘hole’’, among the
It is in fact a basic characteristic of this model. Repeat simulations surrounding stationary spheres. The magnetic moment rotates
with large numbers (usually 2000) of spheres lead to f¼0.33, accordingly.
which represents a mean a of 791 (Eq. 2). The rolling sphere can travel directly to a hole by rolling on
To appreciate the results of the two dimensional (2D) model, one stationary sphere or it can go first to a valley between two
one must accept the assumption that the degree of surface stationary spheres and then down the valley keeping in contact
undulation created during sedimentation is comparable to that with both lower spheres. In the first case, the center of the rolling
created by the falling and rolling of spheres in three dimensions sphere moves in a vertical plane along a great circle of radius r þR,
(3D). This is valid only as a rough approximation. Nevertheless, where r and R are the radii of the rolling and stationary spheres,
our simple and computationally efficient model yields an esti- respectively. When rotating though a valley, R is the radius of the
mate of the rolling angle a for the Griffiths et al. (1960) model. small circle between two respective stationary spheres. In both
Results of the 2D model predict a rather high degree of inclination cases, the center of the rolling sphere travels along a circular
shallowing. It can be expected that the addition of one more sector whose central angle we denote W. The angle of rotation of
degree of freedom in the 3D model could increase a over the 2D the magnetic moment is then
case, thereby producing even lower f values. However, the
situation is not so clear because rotations around non-horizontal a ¼ Wð1þ R=rÞ ð5Þ
axes appear in a full 3D model. The problem cannot therefore be
Because the rotation generally consists of two parts (first to
reduced to the simplified approach in Griffiths et al. (1960), and in
valley, then to hole) whose rotation axes are not identical, we
particular, Eq. (4) for Mr cannot hold true.
cannot simply compute the individual rotation angles and sum
Note that all particles are considered magnetic in the pre-
them up. We must first seek a representative value of f, and by
sented model. Nevertheless, the outcome is the same even if
analogy with the model of Griffiths et al. (1960), a corresponding
magnetic and non-magnetic particles of the same size were
a can be assigned to f by Eq. (2).
randomly mixed. This is valid for all the following models.
We allow the spheres to fall randomly over the stationary
spheres and we determine mean values of the process by
integration. Equal-sized spheres falling on the triangular scheme
yields f ¼0.88. Similar computation for the rectangular scheme

Fig. 3. Shallowing factor (f), observed inclination (IO) and relative magnetic
moment (Mr) simulated by the two dimensional model of randomly settling and
rolling spheres. The horizontal axis indicates the increasing thickness of the
sediment by parameter NL that approximately represents the number of
sedimented layers of spheres. For NL 410, the results become stable and simulated Fig. 4. Basic unit of the triangular scheme. A spherical particle falls (vertical
parameters do not change significantly. In this run, the width of the zone was dotted line), rolls on one sphere and then through the valley between two spheres,
equal to 200r, the total number of spheres was n¼ 2000, and the magnetic field until it reaches a hole. Its magnetic vector rotates during rolling as indicated by
inclination (IF) was 60o. Final values are f¼ 0.33, IO ¼ 30o, and Mr ¼0.35. the arrows.
Author's personal copy

J. Jezek et al. / Computers & Geosciences 49 (2012) 270–277 273

provides f¼ 0.74. Therefore inclination shallowing for equal 4.1.2. Rolling on two spheres (case II)
spheres in both schemes is relatively low. In the case of a smaller If no other spheres are in the vicinity, the rolling sphere may roll
sphere falling on larger spheres, the path to a hole is longer, and along the valley between two stationary spheres, either until a point
Eq. (5) dictates an increase in a, hence tending toward greater where it may start falling again or where it will roll on only one
shallowing. Given sufficiently large differences in radii between sphere. It is also possible (particularly when the centers of stationary
the two spheres, the smaller spheres can rotate 3601 or more. spheres are not at equal heights) that, from the beginning, the
Although it is possible to estimate this effect in the same way as moving sphere only touches one of the stationary spheres and then
above, we do not present the results here. We are more interested rolls on the second stationary sphere. To decide which is the case,
in a more realistic 3D model that describes the process of settling we simulate small rotations around one stationary sphere and see if
where fallen spheres become part of the substrate and change the it causes the moving sphere to encounter the other. It may also
topography of its surface. occur that during rotation on two stationary spheres, the rolling
sphere comes to a point where rotation on only one of the two
spheres remains possible. The transition point is sought and fit by
4. Three dimensional random model numerical iteration; if such a point indeed exists, the sphere rolls
onto it and then we proceed to case I. If not, the rotation proceeds on
Constructing a 3D model of random settling and rolling by both spheres.
simple improvement of the previous 2D model is not easy. The The computation of magnetic vector rotation during rolling on
auxiliary envelope above the stationary spheres becomes a surface two spheres must be solved in small increments (Appendix B). For
on which the center of the rolling sphere should descend in the each increment, the vector rotates around the axis connecting
direction of steepest gradient. This is accomplished by the gradient centers of two stationary spheres (Fig. 5; o-axis) and also around
method of minimalization. Although we constructed such a model, the instantaneous axis connecting the vertex of the cone covering
it was found to be computationally inefficient. This led us to use an the stationary spheres and the center of the rolling sphere (Fig. 5;
approach based on equations of analytic geometry in which we pi-axes). Note that the rotation axis of the magnetic vector is not
repeatedly solve equations of mutual contacts and of rolling on one necessarily horizontal.
or two stationary spheres. We elaborated an algorithm in which we
assign magnetic moments to the particles to calculate the effects of
rolling on the accumulation of the total magnetic moment. 4.1.3. Contact with 3 or more spheres (case III)
When a rolling sphere simultaneously contacts three or more
4.1. Algorithm for settling and rolling of spheres including spheres, we test each pair of spheres by using the same procedure as
magnetic vector rotation described above. If motion in a given direction does not cause a
collision with a stationary sphere, then this direction is a candidate
We generate a random starting position of a new sphere above for continued rolling. The direction with the steepest gradient
the surface of a deposition layer and repeatedly solve the follow- among all possible candidates is chosen, where rolling will proceed.
ing situations until the sphere has found its final position:
4.1.4. Implementation in Matlab
(I) A sphere falls and hits a stationary sphere, or two or more The algorithm was implemented in 64 bit Matlab. To settle
spheres simultaneously. about 30000 spheres, a 4-core PC was left running overnight
(II) A sphere touches one stationary sphere and rolls on it until
reaching another stationary sphere(s) or a point from which
it can fall again.
(III) A sphere touches two stationary spheres and rolls on both, or
only on one of them, until it reaches another stationary
sphere(s) or reaches a point from which it can fall again.
(IV) A sphere touches three or more stationary spheres and either
it is already in a stable position or the position is unstable
and it will continue by case (I) or (II).

During rolling in cases (I) and (II), the magnetic moment is


recorded as the vector rotates. Once the particle’s motion ceases,
a new one is generated.

4.1.1. Rolling on one sphere (case I)


The center of the rolling sphere moves along a great circle path in
a vertical plane as it falls on a stationary sphere. The sphere rolls
(keeping its rotation axis horizontal) to the point where its center is
at equal height with the center of the stationary sphere, from where
it will continue to fall. If another stationary sphere is met before,
rolling is interrupted and will either continue on the newly
encountered sphere or on both spheres (the previous and the newly
found). If more than one stationary sphere is found, the first sphere
that is encountered is chosen. If two spheres are touched simulta- Fig. 5. Scheme of magnetic vector rotation in the case of rolling on two stationary
neously, we go to case II; for three or more spheres, we go to case III. spheres. In this case, the stationary spheres are smaller then the falling sphere. The
In the course of rolling on one sphere, rotation of the magnetic falling sphere is shown in the position where it hit the stationary spheres. Rotation
axes are plotted as dashed (o) and dotted lines (pi, i¼ 1,y,n). An incremental
vector proceeds around a horizontal axis, perpendicular to the plane rotation of the upper sphere consists of small rotations around the o-axis and then
of motion of the rolling sphere’s center. Equations of the magnetic around one of the axes pi. The magnetic vector rotation is shown by the arrows.
vector rotation are in Appendix B. See also the figure in Appendix B.
Author's personal copy

274 J. Jezek et al. / Computers & Geosciences 49 (2012) 270–277

without plotting. The model domain was a square box whose side
dimensions were several tens of sphere radii (Fig. 6). We used
periodic boundary conditions, i.e., a sphere appears on the
opposite side if it rolls out of the model domain.

4.1.5. Basic results


In the course of the simulations, f, a, Io, and Mr evolved similar to
the 2D model (Figs. 3 and 8), reaching stable values after thousands
of spheres were settled. Inclination shallowing is greater (lower f) in
3D than 2D models. For example, settling of equally-sized spheres in
a field inclination of IF ¼601 results in f¼0.10, a ¼ 871, IO ¼9.51, and
Mr ¼0.19. Furthermore, the Mr of 0.19 in the 3D model is 30% lower
than that computed by Eq. (4) (Mr ¼0.27). This is because in 3D, the

Fig. 8. Development of the shallowing factor (f), observed inclination (IO) and
relative magnetic moment (Mr) from the 3D roll-slip model. Results are compared
between the attenuation models AC1 (blue, lower curve), AC2 (red, upper), and
AC3 (black, middle). n is the number of settled equi-dimensional spheres.

rotation of magnetic vectors is not fully described by the Griffiths


et al. (1960) model.

5. The roll-slip model

The 2D and 3D models discussed above are based on the


assumption that the motion of settling spheres on the surface of
stationary spheres takes place solely by rolling. They produce rather
large amounts of inclination shallowing when compared to experi-
mental data. However, at least two effects may restrict the amount
of rolling and reconcile theory with experiments. First, falling
spheres may hit spheres that are already rolling on the substrate,
thus interrupting and/or changing their motion. This effect can
partially be statistically assessed depending on the concentration
of the suspension, which will not be treated here. The second is that
spherical particles may also slide (translate) instead of roll.
Some experimental and theoretical studies have examined the
motion of spherical particles on inclined planes (Smart et al.,
Fig. 6. A sample run of the 3D random sedimentation with rolling (10000 spheres 1993; Zhao et al., 2002) and also their mutual behavior as they
of different sizes). collide when falling in a viscous fluid (Zhao & Davis, 2002).
According to these studies, the motion of the spheres consists of
a combination of rolling and slipping depending on the surface
roughness (relative height of surface inequalities), the coefficient
of friction, and the angle (y), which represents either the slope of
the inclined plane or the angle between the line connecting the
centers of the spheres involved and the vertical axis (i.e., the
instantaneous slope).
Smart et al. (1993) derived equations for a sphere’s rotational
and translational velocity as a function of these parameters. They
found a good agreement between theory and experiment. Surface
roughness appears to be the least important parameter while y is
the most. There is a threshold value (yS) below which particles
roll, and above which particles roll and slip. Both theory and
Fig. 7. The models used for the attenuation coefficient: AC1 - a line fit through the
experiments constrain yS between 121 and 331, depending on the
experimental data on Fig. 9 of Smart et al. (1993), AC2 - an approximation of the coefficient of friction of the material. For instance, experiments
theoretical curve of the same figure, AC3 - a step-wise model. with 140 mm glass spheres yielded yS 251 (Smart et al., 1993),
Author's personal copy

J. Jezek et al. / Computers & Geosciences 49 (2012) 270–277 275

which matches relatively well the theoretical prediction when


using 0.225 for the coefficient of friction (Fig. 9 in Smart et al.,
1993). We use their data to obtain a rolling attenuation coefficient
QS(y) (Fig. 7, models AC1 and AC2). The rotation angle is multi-
plied by this coefficient, which causes a decrease of rotation for
y 4 yS when the attenuation coefficient is less than one. Fig. 7
shows a simple stepwise model (AC3) that, through repeat
simulations, provided similar results (results lying between AC1
and AC2, Fig. 8). The advantage of the AC3 model is that it is
simple to interpret: a sphere makes a full rotation for small y and
for y larger than yS, the sphere half rolls and half slips.

5.1. Basic results–equal spheres

Repeat simulations arrive at a characteristic shallowing factor


of f  0.6 (mean f¼0.613, single standard deviation sf ¼ 0.015) for
a 3D random roll-slip model with equi-dimensional spheres and
using the attenuation coefficient from AC3. Table 1 and Fig. 9
present some of the results. Mr values in the simulations are
only about 5% lower than those computed by Eq. (4) for a given
f (Fig. 9b), which means that the 3D random roll-slip model can be Fig. 10. Comparison of shallowing factors produced by the roll-slip model for
approximately described by only one parameter–the shallowing cases (a) and (b) of simulated unequal dimensional spheres. The horizontal axis is
factor f, as it is in the Griffiths et al. (1960) model. the ratio of the number of large (nL) to the number of small (nS) spheres.

A uniform size distribution (r from 2 to 11), all magnetic.


5.2. Unequal spheres (c)
(d) As in (c), only with small (ro6) spheres magnetic.
We next investigated assemblages of different size compositions: (e) An exponential size distribution (r from 2 to 11, number of
particles decreasing with r).
(a) Large and small spheres (r ¼10 and 5), both magnetic. (f) As in (e), only with small (ro6) spheres magnetic.
(b) As in (a), only with the small spheres (r ¼5) being magnetic.
All simulations used the attenuation coefficient from AC3,
with the moment of the magnetic spheres being proportional to
their volume. When the size distribution of the spheres is
identical, the cases where all spheres are magnetic yield
equivalent results as cases with mixtures of magnetic and
non-magnetic spheres. However, when the size distribution of
magnetic and non-magnetic particles is different, the cases are
not identical. Fig. 10 shows a comparison of cases (a) and (b),
where large and small spheres are mixed in different propor-
tions, starting only with small spheres and then subsequently
adding more and more large spheres. When only small spheres
are magnetic, the net effect is to increase their amount of
rolling, and thus exacerbate inclination shallowing. In cases a, c
and e, when all spheres are magnetic, f is  0.6 (similar to the
case of equally-sized spheres); when only small spheres are
magnetic (cases b, d and f), inclination shallowing is system-
atically exacerbated (Table 2). Repeat simulations show limited
variability and differ from one other only by a few percent.
Nevertheless, the values given in Table 2 should be taken as
Fig. 9. Observed inclination (IO) and relative magnetization (Mr) as a function of characteristic, and not as population means of a given random
field inclination (IF) (data from Table 1). Mre values were computed using Eq. (4). process. Estimating the latter requires an inordinate amount of
computer time.

Table 1
Observed inclination (IO), inclination shallowing parameter (f) and relative
magnetization (Mr) as a function of field inclination (IF) derived from simulations
6. Conclusion
using the 3D random roll-slip model using the AC3 attenuation coefficient. In the
simulations, n equal-sized spheres were settled over a square area whose width is We have developed algorithms to facilitate numerical
60 sphere radii. simulation of rolling and slipping spherical magnetic particles
after they settle though a water column in a random position
IF IO f Mr n
from above. The algorithms allow us to estimate the amount of
80 73.8 0.609 0.437 23324 inclination shallowing produced during deposition of spherical
60 47.7 0.634 0.503 25098 particles under conditions where magnetic interaction and
45 31.9 0.623 0.566 15924 flocculation are negligible. A simplified 2D approach based on
30 19.6 0.615 0.630 24292
10 6.2 0.619 0.679 15319
the model of Griffiths et al. (1960) yields results consistent
with a full 3D model. Assemblages of equal-sized magnetic
Author's personal copy

276 J. Jezek et al. / Computers & Geosciences 49 (2012) 270–277

Table 2
Characteristic values of the shallowing factor (f) for all tested models. Cases (a) and (b) use the results for equal numbers of large and small spheres (see Fig. 10).

model 2D 3D

unrestricted roll-slip unrestricted roll-slip

equal spheres unequal spheres

a b c d e f
f 0.33 0.6 0.10 0.64 0.66 0.25 0.63 0.26 0.62 0.46

spheres produce shallowing factors (f) of  0.6, comparable


to that observed in nature and in laboratory redeposition
experiments. Similar results are obtained for assemblages of
unequal-sized spheres, particularly with uniform and expo-
nential distributions. Inclination shallowing becomes more
pronounced when only the smaller particles are magnetic. This
is relevant to natural conditions, where the remanence-carry-
ing grains are restricted to single domain sizes, which are
commonly smaller than the average grain size of the bulk
sediment. Our numerical results show that rolling of spherical
particles, with or without slip, can significantly contribute to
inclination shallowing, especially in conditions with negligible Fig. A1. Scheme of rolling on two stationary spheres.
magnetic interaction and flocculation. Our emerging experi-
mental results on spherical particles confirm this finding (Jezek
et al., 2010; Bilardello et al., 2011). 910 pages, ISBN 1–58488-291-3):
c ¼ cosðaÞ,s ¼ sinðaÞ
Acknowledgements 2 3
c þ ð1cÞo21 , ð1cÞo1 o2 so3 , ð1cÞo1 o3 þ so2
6 c þ ð1cÞo22 , ð1cÞo2 o3 so1 7
R ¼ 4 ð1cÞo2 o1 þso3 , 5 ðA1Þ
Support was provided by Czech science Foundation, grant
ð1cÞo3 o1 so2 , ð1cÞo3 o2 þ so1 , c þð1cÞo23
GACR 205/09/J028, MSM0021620855 and Deutsche Forschungs-
gemeinschaft (DFG) grant GI712/3-1. Prof. Kieran Mulchrone and (for further use define a function R ¼matrot(o,a))
one anonymous referee, together with editorial handling by Jef Rotated magnetic vector is:
Caers, helped improve the manuscript.
mr ¼ Rm

Appendix A. Supporting information


Appendix B2. Rotation of the magnetic vector when rolling
Supplementary data associated with this article can be found in proceeds on two stationary spheres
the online version at http://dx.doi.org/10.1016/j.cageo.2012.06.013.
Used quantities as in the previous case, and:
X1,Y1,Z1y coordinates of the first stationary sphere centre
Appendix B1. Rotation of the magnetic vector when rolling
X2,Y2,Z2y coordinates of the second stationary sphere centre
proceeds on one stationary sphere
R1 and R2y radii of the two stationary spheres
The centre of rolling sphere (S) rotates around the axis (o)
Used quantities:
connecting the centers of two stationary spheres (S1,S2) while the
x0, y0, z0y initial coordinates of the rolling sphere centre
magnetic vector rotates around the instantaneous axis (p) defined
x, y, zy final coordinates of the rolling sphere centre
by means of Fig. A1. The axis p rotates together with the centre S
X, Y, Zy coordinates of the stationary sphere centre
(see also Fig. 5).
r and Ry radii of the rolling and stationary spheres, respectively
Compute:
my unit magnetic vector attached to the centre of rolling
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sphere
d ¼ ðX 1 X 2 Þ2 þ ðY 1 Y 2 Þ2 þ ðZ 1 Z 2 Þ2
The magnetic vector rotates around a horizontal axis, perpen-
dicular to the plane of motion of the rolling sphere’s center. 2
d þ ðR1 þrÞ2 ðR2 þrÞ2
Compute: e¼
2d
u ¼ ½x0 X, y0 Y, z0 Z,v ¼ ½xX, yY, zZ
q ¼ e=d
o ¼ u  v=9u  v9
xC ¼ X 1 þ qðX 2 X 1 Þ,yC ¼ Y 1 þ qðY 2 Y 1 Þ,zC ¼ Z 1 þ qðZ 2 Z 1 Þ
!
0
uv u ¼ ½x0 xC ,y0 yC ,z0 zC ,v ¼ ½xxC ,yyC ,zzC 
W ¼ arccos ,0 r W r p=2
9u99v9
o ¼ u  v=9u  v9
a ¼ Wð1 þR=rÞ !
uv0
Rotation matrix (p. 349 of Standard Mathematical Tables and W ¼ arccos
Formulae, 31st Edition, D. Zwillinger (ed.), CRC, Boca Raton, 2003, 9u99v9
Author's personal copy

J. Jezek et al. / Computers & Geosciences 49 (2012) 270–277 277

if R1 ¼R2 References
a ¼ Wð1 þR1 =rÞ
Anson, G.L., Kodama, K.P., 1987. Compaction-induced inclination shallowing of the
R ¼matrot(o,a), Eq. (A1) post-depositional remanent magnetization in a synthetic sediment. Geophy-
sical Journal of the Royal Astronomical Society 88 (673–692), 1987.
mr ¼ Rm Arason, P., Levi, S., 1990. Models of inclination shallowing during sediment
compaction. Journal of Geophysical Research 95 (B4), 4481–4499.
else Bilardello, D., Jezek, J., Gilder, S., 2011. Understanding magnetic remanence
n¼90 (or another sufficiently large number of incremental rotations) acquisition through synthetic sediment deposition experiments and numerical
simulations. European Geoscience Union General Assembly, Geophysical
DW ¼ W=n Research Abstracts 13 (EGU2011), 4118.
Blow, R.A., Hamilton, N., 1978. Effects of compaction on the acquisition of detrital
R¼matrot(o, DW), Eq. (A1) remanence magnetization in fine grained sediments. Geophysical Journal of
½x,y,z ¼ ½x0 ,y0 ,z0  the Royal Astronomical Society 52 (13–23), 1978.
Collinson, D.W., 1965. DRM in sediments. Journal of Geophysical Research
70, 4663–4668.
q ¼ R1 =ðR1 þ rÞ Griffiths, D.H., King, R.F., Rees, A.I., Wright, A.E., 1960. Remanent magnetism of
some recent varved sediments. Proceedings of the Royal Society A 256,
xD ¼ X 1 þ qðxX 1 Þ,zD ¼ Z 1 þ qðzZ 1 Þ 359–383.
Jezek, J., Bilardello, D. and Gilder, S., 2010, Understanding magnetic remanence
q ¼ R2 =ðR2 þ rÞ acquisition through synthetic sediment deposition experiments. EOS Transac-
tions AGU, Fall Meeting Supplement, Abstract GP11A-0754.
Jezek, J., Gilder, S.A., 2006. Competition of magnetic and hydrodynamic forces on
xE ¼ X 2 þ qðxX 2 Þ,zE ¼ Z 2 þ qðzZ 2 Þ ellipsoidal particles under shear: Influence of the Earths magnetic field
" # " # on particle alignment in viscous media. Journal of Geophysical Research
xD X 1 xD xE X 2 X 1 xD xE 111 (B12S123).
q ¼ det =det King, R.F., 1955. The remanent magnetism of artificially deposited sediments, Mon.
zD Z 1 zD zE Z 2 Z 1 zD zE
Notic. Roy. Astron. Soc. Geophys. Suppl 7, 115–134.
Nagata, T., 1961. Rock Magnetism. Maruzen, Tokyo 348 pp.
xA ¼ X 1 þ qðX 2 X 1 Þ,yA ¼ Y 1 þ qðY 2 Y 1 Þ,zA ¼ Z 1 þ qðZ 2 Z 1 Þ Shcherbakov, V., Shcherbakova, V., 1983. On the theory of depositional remanent
magnetization in sedimentary rocks. Geophysical Survey 5, 369–380.
Scherbakov, V., Sycheva, N.K., 2010. On the mechanism of formation of deposi-
s ¼ ½X 1 x,Y 1 y,Z 1 z=9½X 1 x,Y 1 y,Z 1 z9 tional remanent magnetization, Geochemistry. Geophysics and Geosystems
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 11, 2, http://dx.doi.org/10.1029/2009GC002830.
Smart, J.R., Beimfohr, S., Leighton, D.T., 1993. Measurement of the translational
r BD ¼ R1 1ðso0 Þ2 and rotational velocities of a noncolloidal sphere rolling down a smooth
inclined plane at low Reynolds number. Physics of Fluids A5, 13–24.
p ¼ ½xxA ,yyA ,zzA =9½xxA ,yyA ,zzA 9 Tarling, D.H., Turner, P., 1999. Palaeomagnetism and Diagnesis in Sediments.
Geological Society of London Special Publications 151, 301.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Tauxe, L., Kent, D.V., 1984. Properties of a detrital remanence carried by hematite
r FD ¼ r 1ðsp0 Þ2 from study of modern river deposits and laboratory redeposition experiments.
Geophysical Journal of the Royal Astronomical Society 77, 543–561.
Tauxe, L., Steindorf, J.L., Harris, A., 2006. Depositional remanent magnetization:
Da ¼ DW ðrBD =rFD Þ sign ðpo0 Þ Toward an improved theoretical and experimental foundation. Earth and
Planetary Science Leters 244, 515–529.
for i¼1:n Torry, E.M., Church, B.H., Tam, M.K., Ranter, M., 1973. Simulated random packing
of equal spheres. Canadian Journal of Chemical Engineering 51, 484–491.
p ¼ q½xxA ,yyA ,zzA =9½xxA ,yyA ,zzA 9
Van Vreumingen, M.J., 1993. The influence of salinity and flocculation upon the
Ri ¼matrot(p,Da), eq. (A1) acquisition of remanent magnetization in some artificial sediments. Geophy-
sical Journal International 114, 607–614.
m ¼ RRi m Verosub, K.L., Ensley, R.A., Ulrick, J.S., 1979. The role of water content in the
magnetization of sediments. Geophysical Research Letters. 6, 226–228.
Weltje, G.J., Alberts, L.J.H., 2011. Packing states of ideal reservoir sands: Insights
½x,y,z ¼ ½xxc ,yyc ,zzc R0 þ ½xc ,yc ,zc 
from simulation of porosity reduction by grain rearrangement. Sedimentary
end Geology 242, 52–64.
Zhao, Y., Galvin, K.P., Davis, R.H., 2002. Motion of a sphere down a rough plane in a
mr ¼ m viscous fluid. International Journal of Multiphase Flow 28, 1787–1800.
Zhao, Y., Davis, R.H., 2002. Interaction of two touching spheres in a viscous fluid.
end Chemical Engineering Science 57, 1997–2006.

Você também pode gostar