You are on page 1of 11

Wear 261 (2006) 856–866

Modeling and simulation of surface roughness in magnetorheological


abrasive flow finishing (MRAFF) process
Sunil Jha, V.K. Jain ∗
Department of Mechanical Engineering, IIT Kanpur 208016, India
Received 1 September 2005; received in revised form 5 January 2006; accepted 24 January 2006
Available online 20 March 2006

Abstract
Magnetorheological abrasive flow finishing (MRAFF) process was developed for super finishing of internal geometries of hard materials. This
process relies for its performance on magnetorheological effect exhibited by carbonyl iron particles along with abrasive particles in non-magnetic
viscoplastic base medium. The extent of finishing action depends on radial and tangential forces coming on abrasive particles due to carbonyl iron
particles (CIPs) arranged in columnar structure in the presence of external magnetic field. Experiments were conducted on stainless steel work
pieces with different combinations of CIP and SiC particles in MRP-fluid for same volume concentration. CIP chain structure and surface roughness
evaluation model have been proposed. Magnitudes of the forces on abrasive particles were then calculated and change in surface roughness was
computed using the model developed to simulate final surface roughness.
© 2006 Elsevier B.V. All rights reserved.

Keywords: MRF; Magnetorheological polishing fluid; MRAFF; Precision finishing

1. Introduction properties of polishing medium, hence adds determinism in con-


trolling surface topography being generated during finishing.
The ultra precision finishing technologies have grown rapidly The present applications of MRF process are limited to flat,
over recent years, and have tremendous impact on the develop- spherical and aspherical surfaces due to dwelling of work piece
ment of new products and materials. With the advent of these in the moving magnetorheological polishing (MRP)-fluid rib-
new materials, manufacturing engineers are facing challenge of bon. To meet the finishing requirements of different geometries
machining and finishing these materials to meet their functional and incorporating better in-process control of finishing forces,
requirements. The available traditional and advanced finishing a hybrid process by combining AFM and MRF was developed
processes alone are incapable of producing desired surface char- [3], and named as magnetorheological abrasive flow finishing
acteristics on complex geometries, and in exercising in-process (MRAFF).
control on finishing action. Abrasive flow machining (AFM) [1] A hydraulically powered MRAFF experimental setup was
process was developed to finish internal complex geometries by designed and fabricated to conduct finishing experiments. A
allowing abrasive laden polymeric medium to flow over it under study was made to understand the effect of magnetic field
pressure. The abrading forces in AFM process are function of strength on reduction in surface roughness (Ra ), and the results
viscosity of viscoelastic polymeric base medium, which is very were reported elsewhere [3]. The role of magnetic field strength
difficult to control during operation. This lacks determinism in on decrease in surface roughness value was clearly observed.
the control of finishing action. In another process developed for After preliminary study, the finishing performance of MRAFF
automated lens finishing, magnetorheological finishing (MRF) process is found to be mainly dependent on MRP-fluid compo-
[2], external magnetic field is used to control the rheological sition for the same magnetic field strength, extrusion pressure
and number of finishing cycles. MRP-fluid composition is one
of the key process parameters affecting final surface roughness
in MRAFF process due to its role in fluid structure formation;
∗ Corresponding author. Tel.: +91 512 2597916; fax: +91 512 2597408. hence it is taken up as a main factor for the present study. In this
E-mail address: vkjain@iitk.ac.in (V.K. Jain). study the effect of size of silicon carbide (SiC) abrasives and

0043-1648/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.wear.2006.01.043
S. Jha, V.K. Jain / Wear 261 (2006) 856–866 857

Nomenclature
a spacing between two abrasive particles passing
over work piece (m)
A total projected area of spherical abrasive grain
(m2 )
A projected area of embedded portion of abrasive in
the work piece (m2 )
B magnetic flux density (T)
Df fixture inner diameter in finishing zone (m)
Dp MRPF cylinder inner diameter (m)
DCIP carbonyl iron particle diameter (m)
Fm magnetic force on CIP in external magnetic field
(N) Fig. 1. Mechanism of MRAFF action.
Fnormal normal force on abrasive particle (N)
Fshear shear force on abrasive particle (N)
H magnetic field strength (A/m) it was compared with the experimental results as discussed in
HBHN Brinell hardness (kgf/mm2 ) the following sections.
I magnetizing current (A)
L length of electromagnet coil (m) 2. Magnetorheological abrasive flow finishing (MRAFF)
Le MRP-fluid extruded (slug) length (m)
Ls MRPF cylinder stroke length (m) Finishing forces in MRAFF process are controlled by rhe-
Lspan CIP chain’s spanning length from one end to ological properties of MRP-fluid which comprises of carbonyl
another (m) iron particles and very fine abrasives dispersed in viscoplas-
m mass of CIP (kg) tic base medium of mineral oil and grease. This composition
n number of turns per unit length (m−1 ) exhibits unique reversible change in its rheological properties
N number of abrasive grains in a line in a stroke on the application and removal of external magnetic field. The
Nv number of particles in volume V of MRP-fluid magnetic field dependent yield stress and viscosity of MRP-
NCIP number of CIP particles in a given MRP-fluid fluid can be controlled by controlling magnetizing current in
r1 core radius of electromagnet coil (m) the electromagnet coils producing magnetic field across the
r2 outer radius of electromagnet coil (m) finishing zone. The CIPs acquire magnetic dipole moment pro-
Rnormal reaction force on abrasive particle due to Fnormal portional to magnetic field strength, and aggregate into chain
Rshear reaction force on abrasive particle due to Fshear like structure aligned in the field direction [4], embedding non-
t depth of indentation (m) magnetic abrasive particles in between (Fig. 1). Depending on
V volume of MRP-fluid (m3 ) the size and volume concentration of abrasives and carbonyl iron
Vu volume of a unit cell (m3 ) particles (CIPs), the bonding strength gained by the abrasives
VCIP volume of a CIP particle (m3 ) through surrounding CIPs chains varies. To finish internal work
x distance from pole face (m) piece surfaces in MRAFF process, the MRP-fluid was extruded
Yi Ordinate of roughness profile data (mm) through the work piece passage in the presence of magnetic field,
as shown in Fig. 1.
Greek letters Abrasion occurs selectively only where the change in rheo-
χm magnetic susceptibility of carbonyl iron particles logical properties of MRP-fluid takes place from near Newtonian
(CIPs) (m3 /kg) to Bingham plastic due to CIPs chain formation. Due to CIPs
φCIP volume fraction of CIPs in MRP-fluid chain formation, non-magnetic abrasive particles get embedded
µ0 magnetic permeability of free space (H m−1 ) between the chains, as shown in Fig. 1, and gain bonding strength
σy yield point stress of stainless steel work piece in in proportion to the magnetic field strength to perform finishing
shear (Pa) action. In this way, the extent of abrasion of peaks by abrasives
τy Fluid shear stress (Pa) is controlled by magnetic field strength and the desired finishing
characteristics are controlled by changing magnetizing current
in the electromagnet.
carbonyl iron particles (CIPs) on decrease in surface roughness
value was investigated on stainless steel work pieces. A micro- 3. Experimentation
scopic study of CIP chain structure formation was conducted.
Based on this observation and suitable assumptions related to Finishing experiments were conducted on a specially
structure formation, surface roughness was simulated using the designed and developed hydraulically powered MRAFF setup
proposed models (chain structures and surface roughness), and [3]. All experiments were conducted for 200 finishing cycles
858 S. Jha, V.K. Jain / Wear 261 (2006) 856–866

experiment nos. 4–6 (Table 1) corresponding to CIPs of 3.5 ␮m


diameter which is much smaller than SiC diameter (7.5–19 ␮m).
Our hypothesis is that the chains formed from small carbonyl
iron particles of HS grade (3.5 ␮m) are not strong enough to hold
the bigger abrasive particles, and are unable to provide required
finishing force. The ability and extent of an abrasive particle to
participate in finishing action in MRAFF process depends on
the normal indentation force and ability of CIPs to hold abrasive
particle. The overall force acting on abrasive particle depends
on volume fraction of CIP and on arrangement of CIPs. In case
of smaller CIPs though more number of particles are present in
Fig. 2. Flow curves of MRP-fluids with 20 vol.% CS (Fluid 1) and HS (Fluid 2) the fluid for same volume fraction but all are not contributing
CIPs with 800 mesh SiC at 2000 G. towards normal force on abrasive particle. Also the magnetic
force on CIPs depends on their mass, which is less in case of
at 3.75 MPa hydraulic extrusion pressure and 0.531 T magnetic smaller particles. The interparticle magnetic force between CIPs
flux density. The MRP-fluid was prepared with 20 vol.% car- which governs the holding force during shear and restrain break-
bonyl iron powder and 20 vol.% silicon carbide abrasives in ing of chains is less in case of smaller particles. This reasoning
60 vol.% base medium of paraffin liquid and AP3 grease. The is supported in the following sections by the theoretical analy-
polishing fluid was prepared by mixing abrasive and iron par- sis of chain structure formation and surface roughness profile
ticles into nearly continuous phase (grease + oil) and stirring simulation.
with the help of specially designed multi-blade mixer. This
results in uniform dispersion of iron and abrasive particles in the 4.1. Chain structure and unit cell modeling
base medium. The flow characteristics of two such preparations
MRPF-20CS-20SiC800 and MRPF-20HS-20SiC800 at 2000 G The study of chain structure formation on the application of
are shown in Fig. 2. The rheological behaviour of MRP-fluids magnetic field helped in understanding the role of particle size on
under magnetic field was evaluated separately using capillary surface finish improvement. The structures of six different MRP-
magnetorheometer and is not discussed here due to different fluid compositions were investigated. To study the magnetic
scope of work. Magnetic field across the work piece was pro- forces acting on abrasive particles during finishing operation,
duced with the help of C-shaped electromagnet, made of cold structural arrangement of carbonyl iron particles around abrasive
rolled annealed steel. Experiments were conducted on stainless particles should be well understood. It is known that the carbonyl
steel ground work pieces (flat surfaces) with initial Ra value of iron particles acquire magnetic dipole moment in the presence
∼0.30 ␮m. The experiments were conducted using two different of external magnetic field. Whenever dipolar interaction forces
grades of carbonyl iron powders, CS and HS (BASF) along with between the particles exceed their thermal interactions, the parti-
black silicon carbide of mesh size 800, 1200, and 2000. Before cles aggregate into chains of dipoles aligned in the field direction
and after every experiment, the surface roughness profiles were [4]. These chains form columnar structure along the magnetic
recorded by Mahr Federal Surfanalyzer 5000. lines of force at higher concentration of carbonyl iron particles.
For simplifying simulation of chain structure, it is assumed that
4. Modeling and simulation iron particles rearrange around SiC particles in the direction
of magnetic field and repeat itself in a complete volume span-
The surface roughness measurement results are summarized ning from one end to the other end of the fixture between the
in Table 1. The change in Ra value was calculated as, Ra = final electromagnet poles. Due to the presence of non-magnetic abra-
Ra − initial Ra . The highest improvement from 0.32 to 0.09 ␮m sive particles in the MRP-fluid, the chains are rarely continuous;
is observed in case of MRP-fluid containing CIP-CS and SiC- instead terminate at the abrasive particles if it comes in between
800 and the least improvement is found in CIP-HS and SiC-2000 the chain path, the same is observed under optical microscope
combination for experiment nos. 1 and 6, respectively. There (Fig. 3). Under the magnetic field, the SiC particles are dispersed
was not much improvement in surface roughness (Ra -value) for in a dense network or structure of interconnected or cross-linked

Table 1
Surface roughness results
Experiment no. CIP dia. (DCIP ) (␮m) SiC dia. (DSiC ) (␮m) DCIP /DSiC Initial Ra (␮m) Final Ra (␮m) Ra a (␮m) %Ra

1 18.0 (CS) 19.00 0.95 0.32 0.09 −0.23 −71.87


2 18.0 (CS) 12.67 1.42 0.28 0.17 −0.11 −39.28
3 18.0 (CS) 7.50 2.40 0.31 0.23 −0.08 −25.80
4 3.5 (HS) 19.00 0.18 0.26 0.23 −0.03 −11.54
5 3.5 (HS) 12.67 0.28 0.28 0.24 −0.04 −14.28
6 3.5 (HS) 7.50 0.47 0.25 0.24 −0.01 −4.00
a Ra = final Ra − initial Ra .
S. Jha, V.K. Jain / Wear 261 (2006) 856–866 859

where n is integer number of SiC in the unit cell, V volume of


MRP-fluid, and NSiC is the number of SiC particles in volume
V. The number of CIPs is calculated that can be accommodated
in this unit cell volume based on the possible equilibrium posi-
tions and space available. Thus, the unit cell volume will be
governed by the number of abrasive particles available in the
medium. Calculations were done to estimate number of parti-
cles per unit cell (volume depends on particle sizes) for particle
sizes specified in Table 1 and the results are summarized in
Table 2. The calculations for Fluid 1 are illustrated in following
Fig. 3. Optical micrograph showing carbonyl iron chain terminating on SiC
abrasive particle (clearly visible in colored image). (For interpretation of the paragraph.
references to color in this figure caption, the reader is referred to the web version Fluid 1 comprises of 20 vol.% CIP (CS grade) of average
of the article.) particle diameter 18 ␮m and 20 vol.% SiC of mesh size 800
with average particle diameter 19 ␮m. Number of iron particles
(NCIP ) in a given volume V of MRP-fluid is given by

φCIP V
NCIP = (2)
VCIP

where φCIP is volume fraction of CIP in MRP-fluid, V the vol-


ume of MRP-fluid, and VCIP is the volume of a single particle. A
Fig. 4. Repetition of unit cubic cell with CIP and SiC arranged in the MRP-fluid 1 mm3 MRP-fluid contains 0.2 mm3 CIP and 0.2 mm3 SiC par-
on application of magnetic field.
ticles. Considering volume of one CIP, the number of CIPs in
1 mm3 is calculated from Eq. (2) and is 65,496. Similarly num-
particulate columns, rather than an isolated columnar structure,
ber of SiC particles in 1 mm3 MRP-fluid is found to be 55,710.
as is visible in Fig. 3. Interconnected structure was also observed
The available volume per SiC particle is 1.795 × 10−5 mm3 .
and reported by Fermigier and Gast [5].
Only one carbonyl iron particle in this volume will be able to get
Iron particles are assumed to be uniform in size and homoge-
accommodated around SiC particle to apply force. Considering a
nously distributed spheres that can be magnetically modeled as
cube around SiC particle of volume 1.795 × 10−5 mm3 , the edge
identical induced dipole moments. The chain is formed by the
of cube comes out to be = 26.18 ␮m. The only possible arrange-
alignment of spherical CIPs in the magnetic field, touching end
ment of CIP of average diameter 18 ␮m in this cube is along
to end diametrically. For non-magnetic silicon carbide abrasive
diagonal of cube as shown in Fig. 5(a). Similarly, following the
particles, it is assumed that there is a little movement of the
above assumptions and procedure, arrangement of CIPs in unit
particles in MRP-fluid on the application of external magnetic
cubic cell around SiC for all other fluid compositions were made
field. They get trapped within the iron chains wherever they were
and the same are shown in Fig. 5. For experiment 4, though there
before the application of magnetic field with slight adjustment,
are 160 CIPs per SiC, the structure shown in Fig. 5(d) shows only
as shown in Fig. 4. So, the number of silicon carbide particles
CIPs exerting force on SiC particle to facilitate understanding.
in unit volume is calculated by considering their uniform dis-
This is based on the assumption that only CIP particles present
tribution inside the MRP-fluid. A unit cubic cell occupies the
around SiC in the cell will exert force on it. Other CIPs are
MRP-fluid volume Vu available around integer number of SiC
arranged in remaining space in cube along magnetic lines of
particles and calculated as
force, and it is assumed that their contribution to the force act-
V ing on an abrasive particle in contact with the work piece is
Vu = n (1)
NSiC negligible.

Table 2
Number of SiC and CIP particles calculated theoretically for six different combinations and size of basic repeating cubic cell
Fluid DCIP (␮m) DSiC (␮m) NCIP /mm3 NSiC /mm3 NCIP /NSiC Cube edge assoc. with
integral number of SiC (␮m)

1 18 19.00 65496 55710 1.18 26.18


2 18 12.67 65496 187803 0.35a 25.20 (3 SiC)
3 18 7.50 65496 905415 0.07b 25.00 (14 SiC)
4 3.5 19.00 8908965 55710 160.00 26.20
5 3.5 12.67 8908965 187803 14.00 18.00
6 3.5 7.50 8908965 905415 10.00 10.30
a NSiC /NCIP = 3.
b NSiC /NCIP = 14.
860 S. Jha, V.K. Jain / Wear 261 (2006) 856–866

Fig. 5. Configuration of carbonyl iron particles and silicon carbide abrasives for six different MRAF-finishing experiments. (a) CIP-CS (18 ␮m), SiC-800 (19 ␮m);
(b) CIP-CS (18 ␮m), SiC-1200 (12.67 ␮m); (c) CIP-CS (18 ␮m), SiC-2000 (7.5 ␮m); (d) CIP-HS (3.5 ␮m), SiC-800 (19 ␮m); (e) CIP-HS (3.5 ␮m), SiC-1200
(12.67 ␮m); (f) CIP-HS (3.5 ␮m), SiC-2000 (7.5 ␮m).

4.2. Analysis of forces particles which are smaller as compared to abrasive parti-
cles. The number of abrasive particles in contact with the
To get an insight into the reasons for variation in reduction in work piece surface reduces significantly when mixed with
surface roughness (Ra ) with different combinations of particle smaller CIPs (experiments 4–6).
sizes of carbonyl iron and silicon carbide, it is required to develop (iv) It is assumed that a spherical abrasive particle penetrates
a mathematical model to predict the forces acting on abrasive into the work piece surface under the action of the normal
particles during MRAFF process. Following assumptions were component of magnetic force on carbonyl iron particles in
made to simplify the analysis and to understand the process the presence of magnetic field.
physics: (v) The MRP-fluid structure assumed at rest is field elongated
chains of particles agglomerates. The effect of fringing
(i) The SiC particles embedded in the carbonyl iron chains field at the corners of the magnetic pole piece is neglected
are assumed to repeat and span from one end to another because of the bigger size of pole piece than the work piece
end of the fixture. This assumption is made to simplify length.
the physical model, though in actual case the chain like (vi) The normal magnetic force on iron particles in a stationary
structures formed between the magnetic poles are more system (no shearing applied), on the application of mag-
complex as can be seen in Fig. 3. Iron chains many times netic field, shows steady rise with time after an initial jump.
terminate at the abrasive particles that come in their way This normal force decreases with strain and reaches plateau
and form clusters by aggregating into cylindrical columns. value at large strains [6]. In calculating normal indentation
(ii) All abrasive particles are assumed spherical of average force this decrease is not considered and the normal force
diameter, calculated from their mesh size number. In prac- is assumed constant, which results in comparatively higher
tice, no two abrasive particles resemble each other in shape indentation.
and size.
(iii) The abrasive particles are assumed uniformly distributed The MRP-fluid structure is deformed under shear stress and
in MRP-fluid. It is observed microscopically also that this its flow is accompanied either by the rupture of chains, their slip
assumption is not very much true in case of carbonyl iron over a wall as a unit or by a combination of these two processes
S. Jha, V.K. Jain / Wear 261 (2006) 856–866 861

Fig. 6. (a) Forces on abrasive particle in MRAFF process and (b) calculation of Di .

[7]. The abrasive grain produces a groove on the work piece where Dg is the abrasive grain diameter in m, Di is the indentation
surface under the action of magnetic force acting on CIPs. This diameter in m, is d = 9.98764 × 10−14 m.
force acting on CIPs is transferred to the abrasive particle(s) Indented abrasive cross-sectional area (Fig. 6), A = 5.26 ×
trapped in the vicinity of CIPs. The cross-section of indentation 10−18 m2 . From Eqs. (3) and (4):
(or groove) corresponds to the profile of the penetrated portion of
the grain. Under hydraulic pressure, when the penetrated abra- Fshear = 1.4885 × 10−5 N; Rshear = 1.78577 × 10−9 N
sive grain is translated horizontally, the removal of work piece
On comparison, the Fshear is found to be much greater than Rshear
material takes place. This occurs only when tangential force by
to perform finishing action.
the fluid (Fshear ) on the projected area of the penetrating abra-
For finishing experiments two multilayered copper coils each
sive (above the portion indented into the surface) is greater than
with 2000 turns of 17 SWG were used to produce magnetic
the reaction force (Rshear ) on the indented projected area of the
field in the gap of 30 mm (Fig. 7). To calculate normal force
abrasive due to the strength of the work piece material as shown
on any ferromagnetic particle (acts as a magnetic dipole) in the
in Fig. 6. The removal of material in front of the abrasive grain
external magnetic field, variation of magnetic flux density (B) is
can takes place either by shearing action (chip formation) or by
calculated between the two pole pieces. Using Eqs. (6)–(8) for
ploughing action, which depends on the depth of indentation
magnetic flux density due to a finite solenoid [9], the value of B
and average cutting edge radius [8]. No evidences (experimen-
at a distance x from the coil 1 (Fig. 7), is the vector sum of B1
tal or theoretical) are available to prove either of the mechanism
by which material is being removed during MRAFF. However,
the forces acting per grain during MRAFF are substantially low.
Hence, to be on safer side, ploughing can be assumed as the
mechanism of material removal. Fnormal is the sum of compo-
nents of forces (normal to the work piece surface) acting on the
abrasive particle due to all CIPs in the unit cell:

Fshear = (A − A )τy (3)

Rshear = A σy (4)

where A is the total projected area of abrasive grain, A projected


area of indented part of abrasive in work piece surface, σ y yield
stress of stainless steel (SS) work piece in shear, and τ y is the
fluid shear stress.
Let us take following data to substitute in Eqs. (3) and (4):

σy = 339.5 MPa for SS;


SiC particle diameter = 19 ␮m(experiment1)

The indentation depth calculated from Eq. (5) derived from the
OAB in Fig. 6(b):
Dg 1 2
d= − Dg − Di2 (5) Fig. 7. (a) Electromagnet configuration for force calculation and (b) fixture
2 2 position in the gap between electromagnet poles.
862 S. Jha, V.K. Jain / Wear 261 (2006) 856–866

where x is distance in meters with reference to coordinate axes


placed at work piece fixture (Fig. 7), and B is magnetic flux
density at a distance x in Tesla. Force on a small ferromagnetic
particle of mass m [10] is given by Eq. (10):

Fm = mµ0 χm H∇H (10)

where µ0 is magnetic permeability of free space, χm magnetic


susceptibility of carbonyl iron particles (CIPs), m mass of CIP,
and H is magnetic field strength. For practical purposes, it is
advantageous to replace field strength with magnetic induction
Fig. 8. Variation of magnetic flux density in the gap. using relation B = ␮0 H, so that Eq. (10) becomes

mχm
(flux density due to coil 1) and B2 (flux density due to coil 2) as Fm = B∇B (11)
µ0
follows.
Hence: Owing to the bigger size of flat magnetic pole piece in com-

B(x)  1 (x) + B
=B  2 (30 − x) (6) parison to the diameter of work piece fixture, the variation of
B in y direction is negligible, hence after neglecting it Eq. (11)
where simplifies to
⎡ 
µ0 In ⎣ r22 + (L + x)2 + r2 mχm dB(x)
 1 (x) =
B (L + x)ln  Fm (x) = B(x) (12)
2(r2 − r1 ) r 2 + (L + x)2 + r µ0 dx
1 1
 ⎤ To calculate the force on an abrasive particle due to magnetic
r22 + x2 + r2 force on carbonyl iron particles near work piece surface, their
−x ln  ⎦ (7)
r1 + x + r1
2 2 structural arrangement and alignment are drawn as per possible
equilibrium structure in unit cell and are shown schematically
in Fig. 5.
µ0 In To evaluate the forces acting on CIPs, mass susceptibility
 2 (30 − x) =
B
2(r2 − r1 ) χm of CS and HS grades is calculated using the M–B curves
⎡  (Fig. 9). Parallel field vibrating sample magnetometer (VSM)
r22 + (L + 30 − x)2 + r2 model 150A was used to find magnetic properties in terms of
× ⎣(L + 30 − x)ln  M–B curve for CIPs. A magnetic field was generated by an elec-
r12 + (L + 30 − x)2 + r1
tromagnet driven by power supply, and a Hall-effect Gauss meter
 ⎤ was used. A cylindrical disc shaped sample of 3 mm diameter
r22 + (30 − x)2 + r2 and 2 mm thickness was prepared and placed in a cylindrical
−(30 − x)ln  ⎦ (8)
glass tube between strong magnet coils of magnetometer. When
r1 + (30 − x) + r1
2 2
the mechanical vibrations were applied to the magnetic material
where r1 , r2 and L are core radius, outer radius and length of in a constant magnetic field, the induced voltage proportional
electromagnet coil, respectively. x is the distance from pole face to the magnetic moment of the material was generated in the
(Fig. 7(a)), µ0 is magnetic permeability of free space, I is mag- pick-up coils. These are plotted to obtain curves for CS and HS
netizing current, and n is number of turns per unit length. The CIP grades as shown in Fig. 9.
variation of flux density in the air gap due to electromagnet was To understand the arrangement of SiC and CIPs in a unit cell,
also experimentally measured using a Gauss meter. The mea- let us consider Fluid 1. From Table 2, it is found that there are
surements were done from coil 1 to coil 2 in forward direction ∼1.18 carbonyl iron particles per abrasive particle in unit cell
and then from coil 2 to coil 1 in backward direction. Both these volume. With the abrasive particles in contact with the work
readings were almost overlapping so average is taken and the piece top surface the possible arrangement of one CIP around
value of B is plotted a ‘*’ along with theoretical variation as one SiC particle in a cube of 26.18 ␮m edges is shown in Fig. 5(a)
solid line in Fig. 8. The relative permeability (µr ) of iron core which fits along diagonal direction only. Magnetic force Fm on
was found out by measuring B in the presence and absence of the CIP (φ = 18 ␮m) at a distance of 20.18 ␮m from the work
iron core and it was 7.26. The theoretical variation of B in the gap piece surface, and at x = 3.02018 mm (3 mm is fixture and work
is quadratic, therefore after fitting the quadratic equation (dashed piece thickness +0.02018 mm is perpendicular distance of CIP
curve in Fig. 8) to the experimentally obtained variation of B, from the work piece surface) is calculated as follows from Eq.
we get (9):

B(x) = 203.22x2 − 5.0778x + 0.5711 (9) B = 0.557618 T


S. Jha, V.K. Jain / Wear 261 (2006) 856–866 863

Fig. 10. Cutting mechanism by abrasive particles surrounded by CIP clusters.

through the fixture cross-section can be evaluated as follows:


πDp2 Ls πDf2 Le
=
4 4

Dp2 Ls
Le = (14)
Df2
where Dp and Ls are MRPF-cylinder piston diameter and stroke
length, respectively. Df is the fixture’s inside diameter. The num-
ber of abrasive particles in an extrusion length of 459.201 mm
passing over the peaks in a line are calculated from Le /a, where
a is the edge of cube around each abrasive particle or linear
spacing between the center of two adjacent abrasive particles or
clusters (Fig. 10). Number of abrasive particles passing over a
roughness peak per stroke is equal to the number of abrasive par-
Fig. 9. M–B curve for CIP of CS and HS samples. ticles or multiple thereof (if more than one abrasive in unit cell)
in the extruded length, and is shown in last column of Table 3.

Differentiating Eq. (9) with respect to x and substituting the 4.3. Surface roughness simulation
value of x = 0.003028, we get
Initial surface roughness data input to the model was taken
dB from the Surfanalyser 5000 instrument in the form of ordinates
= −3.85028
dx of all sampled points in the profile at equal interval. To update
The value of χm at B = 0.557618 T obtained from B–M curve the surface profile after each stroke, depth of indentation (d) by
using Eq. (13) is χm = 3.985 × 10−4 m3 /kg: a spherical abrasive on each peak is calculated from Eq. (5). The
initial heights or depths on the surface are shown in Fig. 11 (Ya ,
M µ0 M Yb , Yc , Yd , and Ye ,). The new peak heights updated after one
χm = = (13)
H B indentation depth (or cutting by one grain) are calculated as Ya ,
Yb = Yb − d, Yc = Yc − d, Yd and Ye = Ye − d. The indenta-
Substituting these values in Eq. (12), we get Fm = 16.216 ×
tion diameter Di can be calculated from Brinell hardness number
10−9 N. Following the above procedure, forces for other cases
(kgf/mm2 ) [11] using Eq. (15) given below:
also have been computed, and the same are given in Table 3.
The number of abrasive particles taking part in work piece 2F
HBHN =    (15)
finishing action during a single stroke can be calculated based on
πDg Dg − Dg2 − Di2
volume constancy of the MRP-fluid. The extrusion length (Le )

Table 3
Calculated force and depth of indentation
Experiment no. CIP dia. SiC dia. Force/SiC particle Depth of indentation No. of abrasives in
(DCIP ) (␮m) (DSiC ) (␮m) (×10−9 N) (×10−14 m) one strokea

1 18.0 19.00 16.279 9.987 17540


2 18.0 12.67 5.406 4.997 27333
3 18.0 7.50 2.318 3.620 36736
4 3.5 19.00 7.032 4.335 17540
5 3.5 12.67 1.614 1.492 25511
6 3.5 7.50 0.232 0.372 44583
a Only those abrasives which are passing over the peaks in a line.
864 S. Jha, V.K. Jain / Wear 261 (2006) 856–866

Fig. 11. (a) Abrasive grain approaching initial peaks/valleys of height/depth and (b) new peak heights updated after one indentation depth (d).

Rearranging Eq. (15) gives where Fi is normal indenting force on abrasive in N, and HBHN
 is the work piece Brinell hardness number in kgf/mm2 . Let us
2
2 × 10−6 Fi assume that each abrasive grain passing over the peaks penetrates

Di = Dg − Dg −
2 (16)
9.81HBHN πDg equal to d and each abrasive grain is active. The new peak height

Fig. 12. Flow chart for surface roughness profile simulation in MRAFF.
S. Jha, V.K. Jain / Wear 261 (2006) 856–866 865

Yi after one stroke with Ng active grains/stroke is given by Table 4


Comparison of experimental and theoretically simulated surface roughness val-
Yi = Yi − Ng d (17) ues after 200 finishing cycles
Experiment CIP dia. SiC dia. Initial Final Ra (␮m) %Error
If the point is not a peak point in the data file (for example Ya no. (DCIP ) (DSiC ) Ra (␮m)
and Yd in Fig. 11), then it is transferred to next profile as it is. (␮m) (␮m) Expt. Theor.
To update the peak heights, new peak points were calculated
1 18 19 0.32 0.09 0.21 +133.33
after each stroke and updated profile data were passed on for 2 18 12.67 0.28 0.17 0.19 +11.76
processing in the next stroke. Before and after each stroke, the 3 18 7.5 0.31 0.23 0.24 +4.35
center-line-average (CLA) surface roughness value (Ra ) from 4 3.5 19 0.26 0.23 0.21 −8.69
profile data points was calculated using Eq. (18): 5 3.5 12.67 0.28 0.24 0.25 +4.17
6 3.5 7.5 0.25 0.24 0.23 −4.17

|Yi |
Ra = i=1,...,N (18)
N marized in Table 4. These theoretically calculated Ra values are
where N is number of data points and Yi is roughness pro- in close agreement with the experimental results with maximum
file height at the data points. To calculate final Ra value based error of 12% except for experiment 1. In experiment 1, the error
on theoretical indentation values, software in ‘C’ programming is high due to assumption that MRP-fluid is in sheared state and
language was written as per flow chart shown in Fig. 12. The only one CIP is contributing force on abrasive particle. The-
theoretically obtained final Ra values after simulation are sum- oretically, force contribution on one SiC particle by only one

Fig. 13. (a) Initial input profile, (b) final simulated profile and (c) measured final profile after 200 cycles for experiment 2 with DCIP = 18 ␮m and DSiC = 12.67 ␮m.

Fig. 14. (a) Initial input profile, (b) final simulated profile and (c) measured final profile after 200 cycles for experiment 5 with DCIP = 3.5 ␮m and DSiC = 12.67 ␮m.
866 S. Jha, V.K. Jain / Wear 261 (2006) 856–866

CIP was considered (CIP/SiC = 1.18) though in actual experi- breaking of chains in case of smaller particles. The abrasive
mental conditions neighboring CIPs will also exert force. Due particles in these cases are too large to be effectively cap-
to contribution from more than one CIPs towards the total nor- tured by the CIP since it is easier to shear or slide through
mal indenting force on abrasive particle, the total indentations them due to lack of sufficient bonding force.
obtained experimentally were higher than theoretically calcu- (iii) The theoretically obtained profile for all experiments
lated one and hence the error was more. This is prominent in except experiment 1, matches to a large extent with the
case where size of CIP and SiC is almost equal and the structure measured final profiles. This shows a close resemblance of
was more closely packed. For experiments 2 and 5, the simulated simulated model with the process physics and mechanism
and actual profiles are shown in Figs. 13 and 14. Comparatively of finishing action.
large indentations are obtained theoretically when the abrasive
particles and CIPs are of approximately same size (experiment 1) Acknowledgements
as compared to the case when CIPs are smaller in size compared
to abrasive particles, say, in experiments 4–6. This conclusion We sincerely thank BASF Germany for arranging carbonyl
closely resembles with the experimental observations. Thus it iron powders of different grades for our research work. We
can be concluded that the theoretical modeling and simulation acknowledge the Council of Scientific and Industrial Research
clearly captures the fact that CIP and abrasive particle size ratio (CSIR), New Delhi, India for their financial support for project
plays an important role in final roughness value obtained after no. 5411/NS/02/EMRII entitled “Magnetorheological Abrasive
MRAFF process. Flow Finishing (MRAFF)”. We also acknowledge financial sup-
port of Department of Science and Technology (DST), New
5. Conclusions Delhi, under DST-NSF Indo-US S&T cooperation program vide
project no. DST-INT-US (NSF-PRO-101)/2002.
A model for the CIP chain structures around abrasive parti-
cle has been proposed based on microscopic study and possible References
arrangement of particles with different size in MRP-fluid com-
position. Final surface roughness from initial profile data and [1] L.J. Rhodes, Abrasive flow machining: a case study, J. Mater. Process.
proposed surface roughness model has been simulated for all Technol. 28 (1991) 107–116.
[2] W.I. Kordonski, S.D. Jacobs, Magnetorheological finishing, Int. J. Mod.
combinations of SiC and CIP sizes. Following are the main con- Phys. B 10 (23–24) (1996) 2837–2848.
clusions made after experimental study, theoretical modeling, [3] S. Jha, V.K. Jain, Design and development of magnetorheological abrasive
and simulation: flow finishing (MRAFF) process, Int. J. Mach. Tool Manuf. 44/10 (2004)
1019–1029.
(i) For the same magnetic flux density, the finishing forces on [4] E.M. Furst, A.P. Gast, Micromechanics of magnetorheological suspen-
sions, Phys. Rev. E61/6 (2000) 6732–6739.
abrasive particles are mainly dependent on number of CIPs [5] M. Fermigier, A.P. Gast, Structure evaluation in a paramagnetic latex sus-
in their vicinity, their microstructural arrangement and size. pension, J. Colloid Interf. Sci. 154 (1992) 522–539.
The magnetic force on a carbonyl iron particle is a func- [6] S. Howard, R. Tanner, Shear rate dependence of the normal force of a
tion of particle volume, so size of CIP in comparison with magnetorheological suspension, Rheol. Acta 42 (1–2) (2003) 166–170.
abrasive size is an important factor affecting final surface [7] S. Gorodkin, N. Zhuravski, Surface shear stress enhancement under MR
fluid deformation, Int. J. Mod. Phys. 16 (17–18) (2002) 2745–2750.
roughness obtained in MRAFF process. [8] J.N. Brecker, R. Brown, T. Matsuo, K. Saito, J.A. Sweeney, J.B. Vansaun,
(ii) Compared to the same size of CIP and abrasive particle, the M.C. Shaw, Fourth Annual Report of Abrasive Grain Association on Inves-
surface finish improvement rate decreases with decrease in tigation of Abrasive Grain Characteristics, Carnegie Institute of Technol-
abrasive particles size (keeping CIP size constant) due to ogy, USA, November 1969.
decrease in indenting force and sharing of the same force by [9] www.netdenizen.com/emagnet/solenoids/solenoidonaxis.htm.
[10] A.W. Stradling, The physics of open-gradient dry magnetic separation, Int.
more number of abrasive particles. This is also because of J. Miner. Process. 39 (1993) 19–29.
decrease in the interparticle magnetic force between CIPs [11] H.W. Hayden, W.G. Moffatt, J. Wulff, The Structure and Properties of
which governs the holding force during shear and restrain Materials, vol. III, John Wiley and Sons, 1965.

You might also like