Abstract This paper presents a review of the reportedresearch investigations that are related to the solid particleerosion behavior of polymers and polymeric composites.Attention is paid to the effects of test parameters such aserodent type, size of the erodent, impingement angle, impactvelocity and stand of distance. On the erosion wear rate ofpolymer composites. Various predictions and models pro-posed by different authors to describe and quantify theerosion rate are discussed and their suitability is checked.Recent findings on the erosion response of multi-componenthybrid composites are also presented. Lastly the implemen-tation of the design of experiments and statistical techniquesin making a parametric appraisal of the erosion processes ofcomposites is discussed.
Solid particle erosion (SPE) wear, which results from thesolid particles moving at various velocities and impinge-ment angles striking a surface of a material is one of themost encountered modes of wear and has recently been asubject of a number of researches. SPE is the removal ofmaterial from the surface by the repeated impact of hard
R. Kaundal (�)Department of Mechanical Engineering, Jawahar Lal NehruGovernment Engineering College Sundernagar, Distt. Mandi, HP175018, Indiae-mail: [email protected]
and angular particles traveling at considerable velocities.It has been recognized as a serious problem in variousengineering applications, including pipelines, valves han-dling gases, hydraulic systems, aerospace components andliquid impellers. Various ferrous and non-ferrous materi-als are extensively used in erosive wear situations. Hencesolid particle erosion of surfaces has received consider-able attention in the past decades. Erosion studies havebeen conducted for a variety of reasons and an exhaus-tive data-base regarding the effect of impact related (impactvelocity and impingement angle), particles related (par-ticle hardness, size, shape and friability) and materialrelated (hardness, ductility and microstructure) variableson the erosion behaviour of metals and alloys is alreadyavailable in the literature. Many have attempted to cor-relate the erosion rate with such variables. Finnie [1]after 40 years involvement with erosion presented in1995 an article on the past and the future of erosion.In this article, the influencing parameters and dominatingmechanisms during solid particle erosion were reviewedon the erosion response of metals and ceramic materi-als. In the same year, another article was published byMeng et al. [2] providing information about the exist-ing wear models and prediction equations. This articlewas more general as it discussed all the frictional phe-nomena termed to ‘wear’ including also the solid particleerosion.
Polymer composites acquire an important place when itcomes to operating in a dusty environment where resistanceto erosion becomes an important feature. Thus, the subjectof erosion wear of polymer composites has been receiv-ing substantial research attention since past few decades.Interest in this area is commensurate with the increasing uti-lization of composites in aerospace, transportation and pro-cess industries, in which they can be subjected to multiple
solid or liquid particle impact. Examples of these appli-cations are pipe lines carrying sand slurries in petroleumrefining, helicopter rotor blades [3], pump impeller blades,high speed vehicles and aircrafts operating in desert envi-ronments, water turbines, aircraft engines [4], missile com-ponents, canopies, radomes, wind screens [5] and outerspace applications [6]. Resistance to rain and sand erosionis called among the major issues in the defence applicationof non-metallic materials [5].
2 Variables Affecting Solid Particle Erosion
The material removal during erosion is dependent on manyinterrelated factors that include the properties and structuresof the target material, the macro- and micro-exposure con-ditions and the physical and chemical characteristics of theerodent particles. The combination of all these factors some-times exceeding 20 in number, results in erosion rates thatare peculiar to specific sets of conditions. Additional diffi-culties arise from the fact that the different processes occursimultaneously during erosion. For example, a low impactangle favours wear processes similar to abrasion because theparticles tend to track across the worn surface after impact.A high impact angle causes wear mechanisms which aretypical of solid particle erosion.
In general, various factors, which influence the erosivewear performance of polymers and their composites, areshown in Fig. 1.
Erosive Wear* Mechanism* Wear rate
Properties of target material* Composition*Microstructure *Fiber/Filler/Matrix interface*Mechanical property profile*Amount of fiber/filler*Fiber orientation*Surface Roughness
Erodent Properties*Shape*Size*Hardness*Type of erodent*Particle feed rate* Erodent Temperature
Experimental Conditions*Angle of impingement*Impact Velocity*Temperature*Particle flux (mass per time)*Erosion Media (air or water)
Fig. 1 Factors affecting the erosive wear performance of polymersand their composites
3 Erosion Wear of Polymer Composites
Polymers and their composites have generated wide inter-est in various engineering fields, particularly in aerospaceapplications, in view of their good strength and low densityas compared to monolithic metal alloys. There have beenvarious reports of applications of polymers and their com-posites in erosive wear situations in the literature [3, 4, 7].Available reports on the research work carried out on ero-sion can be classified under three categories: experimentalinvestigations, erosion model developments, and numericalsimulations. It was recognized quite early that correlation isneeded between the experimental conditions and the erosiveresponse of the tested materials. Tilly [8] presented a thor-ough analysis of the various parameters affecting erosion,including particle properties, impact parameters, particleconcentration, material temperature and tensile stress. Healso reviewed the different mechanisms of erosion, whichwere categorized into brittle and ductile behaviors. Ruffand Wiederhorn [9] presented another review of the solidparticle erosion phenomena considering single and multi-ple particle models on erosion of metals and ceramics. Thesignificant parameters for eroding particles and materialcharacteristics were also presented. Humphrey [10] reporteda comprehensive review of the fundamentals of fluid motionand erosion by solid particles. The review includes a dis-cussion of the experimental techniques and the variousfundamental considerations relating to the motion of solidparticles. An assessment of the fluid mechanics phenomenathat can significantly influence erosion of material surfacesby impinging particles was also presented.
After developing primitive fiber reinforced plastics(FRP) in 1940s, they have been widely used becauseof their superior specific strength and also high corro-sion resistance. Following the development of these high-performance fibres, use of FRP into industrial applicationssuch as load bearing parts of buildings, bridges, tank/vesselsand transportations can be recognized [11, 12]. To ensurethe durability of FRPs for industrial applications, it isnecessary to discuss the degradation behaviour and mech-anism under various conditions such as stress, corrosionand erosion etc. Several parts and equipments are exposedto erosive conditions, for example, pipes for hydraulicor pneumatic transportation [13–15], nozzle and impellerfor sand-blasting facility [16], internal surface of vesselsused for fluidized bed or with catalysis [17–19], nose ofhigh-velocity vehicle [20], blades/propellers of planes andhelicopters [21] etc., some of them made from fibrouscomposites.
The most important factor for design of compositesis the fibre/filler content, as it controls the mechanicaland thermo-mechanical properties. In order to obtain thedesired material properties for a particular application, it is
Silicon (2014) 6:5–20 7
important to know how the material performance changeswith the fibre content under given loading conditions. Thesolid particle erosion behaviour of polymer composites asa function of fibre content has been studied to a limitedextent [22, 23]. Polymer composites with both discon-tinuous and continuous fibre reinforcement possess usu-ally very high specific (i.e. density related) stiffness andstrength when measured in plane. Therefore, such compos-ites are frequently used in engineering parts in automobile,aerospace, marine and energetic applications. The highspecific strength and stiffness of polymers are primarilyresponsible for their popularity. However, the resistance ofpolymers to solid particle erosion has been found to be verypoor [24], and in fact it is two or three orders of magni-tude lower than metallic materials [25]. One possible way toovercome such a shortcoming is to introduce a hard secondphase in the polymer to form polymer matrix composites(PMCs). A number of investigators [24–31] have evaluatedthe resistance of various types of PMCs to solid particleerosion. Tilly [24] and Tilly and Sage [26] tested nylonand epoxy reinforced with various fibres such as graphite,glass and steel and concluded that the reinforcement caneither increase or decrease the erosion resistance depend-ing on the type of fibres. Zahavi and Schmitt [25] tested anumber of PMCs for erosion resistance and concluded thatglass-reinforced epoxy composite had a particularly gooderosion resistance. Pool et al. [3] conducted erosion testson four PMCs and inferred that wee-handled, ductile fibresin a thermoplastic matrix exhibit the lowest erosion rates.The above study was extended further by Tsiang [27]. Hecarried out sand erosion tests on a wide range of thermosetand thermoplastic PMCs having glass, graphite and Kevlarfibres in the forms of tape, fabric and chopped mat as rein-forcements. Kevlar fibres in an epoxy resin provided thebest erosion resistance. In a recent study, Mathias et al.[28] and also Karasek et al. [30] have evaluated the erosionbehaviour of a graphite-fibre-reinforced bismaleimide poly-mer composite. These investigators observed the erosionrates of the PMC to be higher than the unreinforced poly-mer. Many of the investigators [25–28, 31] also consistentlynoted that the erosion rates of the PMCs were considerablylarger than those obtained in metallic materials. Researchon erosion of polymer composites started aggressively afterthe year 2000 and before that only a few studies on thesubject were reported. Table 1 presents the summary ofsome important research findings before year 2000. Mostof these papers have provided only the experimental find-ings and reports on statistical interpretation of the resultswere rare. However, in later years, a number of researchersinvestigated on the erosion wear characteristics of poly-mer composites with both experimental and modelingapproaches. Table 2 presents an overview of these investiga-tions reported after year 2000 on the issues related to solid
particle erosion of fiber and particulate reinforced polymercomposites.
The information presented in the tables is indicative ofthe fact that there has been a growing impetus on erosionresearch after the year 2000. A close study of these workswould reveal certain characteristic trends in the erosionresponse of polymer composites which can be summarizedas follows:
• The fiber loading, stand-off distance, erodent size,impingement angle and impact velocity are the sig-nificant factors in a declining sequence affecting theerosion wear rate. Although the effect of impact veloc-ity is less compared to other factors, it cannot be ignoredbecause it shows significant interaction with stand-offdistance, fiber loading and erodent size. Impact energyof the particle increased with velocity which leads to anincrease in erosion rates.
• Polymer matrix composites exhibit a semi-ductile ero-sion behavior, that is, the erosion rate is maximum atimpact angles between 45◦ and 60◦ , while in certainother studies [23, 71, 72] it has been reported that thereis a pure ductile erosion response with a maximumerosion rate at 30◦ impact.
• At the same time, it must be stated that for most of thepolymer matrix composites the velocity exponents arefound to be between 1.5 and to 2.9 [34, 36, 73]. The ero-sion response of materials is determined by the angulardependence of the erosion rate.
• It was observed that the effect of fibre orientation onthe erosive wear decreases as the impact angle movestowards 90◦ because the fibre shape does not changedramatically with change in fibre orientation [23]. Theerosion rate of composites is higher in the case ofperpendicular impact (90/90) than the case of parallelimpact (90/0).
4 Effect of Experimental Conditions on Erosion Wear
4.1 Effect of Impact Velocity
The impact velocity (υ) of the erosive particle with whichit strikes the target surface has a very strong effect on thewear process. If the velocity is very low then stresses atimpact are insufficient for plastic deformation to occur andwear proceeds by surface fatigue [74]. When the velocityincreases, the eroded material may deform plastically uponparticle impact. In this regime, wear is caused by repetitiveplastic deformation. At brittle wear response, wear proceedsby micro-cutting and subsurface cracking. At very high par-ticle velocities melting of the impacted surface may even
8 Silicon (2014) 6:5–20
Tabl
e1
An
over
view
ofre
sear
chon
the
eros
ion
ofpo
lym
er/p
olym
erm
atri
xco
mpo
site
san
dth
eke
yis
sues
asre
port
edby
inve
stig
ator
sbe
fore
the
year
2000
Mat
rix
Fibe
r/Fi
ller/
Vf
/Wf
%E
rode
ntE
rode
ntsh
ape
Ang
leV
eloc
ityK
eyIs
sues
Yea
rof
Ref
eren
ce
mat
eria
lan
dsi
ze(μ
m)
(deg
)(m
/s)
publ
icat
ion
Nyl
on/E
P30
%,7
0%
(wf)
GF,
Qua
rtz
Irre
gula
r,90
100–
1,00
0ft
/sE
ffec
tof
inte
ract
ion
1970
[26]
25%
(wf)
CF,
80%
(wf)
125–
150
ofpa
rtic
lean
dm
ater
ial
stee
lpow
der
beha
viou
ron
the
eros
ion
rate
.
PA/E
P/PP
S40
%(w
f)
chop
ped
Silic
aSa
ndSp
heri
cal,
155
30,4
5,31
Eff
ecto
fty
peof
fibe
r19
86[3
]
glas
sfi
ber,
Con
tinuo
usG
F,60
,90
and
fibe
rlo
adin
gon
the
Wov
enG
F,an
dW
oven
CF
eros
ion
wea
r
Bis
mal
eim
ide
20%
,30
%,4
6%
,A
l 2O
3A
ngul
ar,4
2,90
60E
ffec
tof
the
erod
ents
ize
1991
[32]
60%
(wf)
Bis
phen
ol63
,143
,390
(im
pact
ing
part
icle
s)
onth
eer
osio
nra
teof
bism
alei
mid
epo
lym
er
EP/
Phen
olic
/E
-Gla
sspl
ain
Silic
asa
ndA
ngul
ar,
30,9
038
,45
Eff
ecto
fim
pact
angl
e19
94[3
3]
PET
and
wea
ve15
0–25
0an
dim
pact
velo
city
wov
enfa
bric
onth
ebe
havi
our
of
poly
mer
mat
rix.
EP
65%
(wf)
trea
ted
CF,
SiC
abra
sive
sIr
regu
lar,
15,3
0,
61.4
%(w
f)
CF
100–
150
45,6
0,90
19,3
5,56
Eff
ecto
fin
terf
acia
l19
96[3
4]
stre
ngth
onth
e
eros
ion
ofFR
Ps.
Silicon (2014) 6:5–20 9
Tabl
e2
An
over
view
ofth
eer
osio
nst
udie
spe
rfor
med
onpo
lym
erm
atri
xco
mpo
site
saf
ter
the
year
2000
Mat
rix
Fibe
r/Fi
ller/
Vf
/Wf
%E
rode
ntm
ater
ial
Ero
dent
size
(μm
)A
ngle
(deg
)V
eloc
ity(m
/s)
Yea
rof
Ref
eren
ce
publ
icat
ion
EP
68%
(vf)
glas
sfi
ber,
glas
sfi
ber
(mod
ifie
d)C
orun
dum
part
icle
s60
–120
30,6
0,90
7020
00[3
5]
PEE
K65
%C
F,(V
f),
unid
irec
tiona
lfib
ers
Stee
lbal
ls30
0–50
015
,30,
45,6
0,75
,90
45,8
520
02[3
6]
PP40
–60
%(V
f),
glas
sfi
ber
Cor
undu
mpa
rtic
les
60–1
2030
,60,
9075
2002
[23]
PEE
K,P
EK
,PE
KK
0%
,20
%G
F,30
%G
F,(1
0%
CF
Silic
aSa
nd15
0–21
215
,30,
60,9
030
,68,
9020
03[3
7]
+10
%PT
FE+1
0%
Gra
phite
)
(Wf),
shor
tfib
ers.
EP
56%
CF,
53%
GF,
(Vf),
unid
irec
tiona
lfib
ers
Stee
lbal
ls30
0–50
015
,30,
45,6
0,75
,90
4520
03[3
8]
PUR
Al 2
O3(0
–64
%),
(Wf)
SiO
240
–70
4524
.820
05[3
9]
Res
inM
atG
F(9
.4%
,17.
1%
,24.
5%
),C
loth
GF
(12
%,
Cra
shed
glas
spo
wde
r35
020
–90
24.5
2006
[40]
27.9
%,3
2.4
%),
UD
GF
27.8
%,(
Vf),
chop
ped
stra
ndm
ats,
plai
nw
eave
,uni
dire
ctio
nal(
UD
)
EP
55.8
%A
F,53
.8%
PBO
,(V
f),
cros
spl
ySi
C10
0–15
015
–90
57.8
2006
[41]
EP,
EP
+Fly
ash
(1:4
)G
F,cr
oss
ply
Silic
aSa
nd15
0–25
030
,45,
60,9
024
,35,
5220
06[4
2]
EP
46.5
%(V
f)
cros
spl
yG
F,1–
4%
Silic
asa
nd15
0–25
030
,45,
60,9
024
,35,
5220
06[4
3]
(wf)
whe
atfl
our
PPS
40%
GF
+25
%C
aCO
3,(
Wf),
shor
tfib
erSi
lica
Sand
150–
200
15,3
0,45
,60,
75,9
020
,40,
6020
07[4
4]
EP
55%
GF,
(Vf),
[45/
-45/
0/45
/-45
/0]s
SiC
400–
500
30,6
0,90
42.5
2007
[45]
PEI
40%
CF,
(Vf),
plai
nw
eave
Silic
aSa
nd10
6–12
015
,30,
45,6
0,75
,90
26.8
820
07[4
6]
PEE
KC
F,un
idir
ectio
nalf
iber
sA
rizo
naTe
stD
ust,
10,1
0015
,30,
45,6
0,90
61,9
7.5,
152.
420
07[4
7]
Siev
edR
unw
aySa
nd
PEI
CF,
unid
irec
tiona
lfib
ers
Silic
aSa
nd15
0–20
015
,30,
45,6
0,75
,90
1.96
,2.8
820
07[4
8]
PEI
0%
,20
%G
F,30
%G
F,40
%G
F,25
%C
F,25
%G
FSi
lica
Sand
150–
300
15,3
0,60
,90
30,5
2,60
,88
2007
[49]
+15
%PT
FE+
15%
(MoS
2+
grap
hite
),sh
ort-
fibe
rs.
PPS
51%
CF,
(Vf),
cros
spl
ySi
lica
Sand
150–
200
15–9
020
,40,
6020
08[5
0]
PEI
60%
CF,
(Vf),
unid
irec
tiona
lfib
ers
Silic
aSa
nd15
0–25
015
,30,
60,9
025
–66
2008
[51]
EP
55%
GF,
(Vf),
[45/
-45/
0/45
/-45
/0]s
SiC
400–
500
30,6
0,90
42.5
2008
[52]
EP
66.4
%G
F,59
.4%
GF,
64%
CF,
(Vf),
Silic
aSa
nd15
0–20
090
25,3
7,47
,60
2008
[53]
uni&
bidi
rect
iona
lfib
ers
PET
30%
GF,
40%
GF,
50%
GF,
(Wf),
cros
spl
ySi
lica
Sand
300,
500,
800
30,6
0,90
32,4
5,58
2008
[54]
PET
50%
GF
+0
%,1
0%
,20
%A
lum
ina,
(Wf),
cros
spl
ySi
lica
Sand
300,
500,
800
45,6
0,90
32,4
5,58
2008
[55]
PET
50%
GF
+0
%,1
0%
,20
%Si
C,(
Wf),
cros
spl
ySi
lica
Sand
300,
500,
800
45,6
0,90
32,4
5,58
2008
[56]
PET
30%
GF,
40%
GF,
50%
GF,
(Wf),
cros
spl
ySi
lica
Sand
300,
500,
800
30,6
0,90
32,4
5,58
2008
[57]
EP
Uni
dire
ctio
nala
ndm
ultid
irec
tiona
lcar
bon
fibe
rSi
C80
15–9
070
2009
[58]
10 Silicon (2014) 6:5–20
Tabl
e2
(con
tinue
d)
Mat
rix
Fibe
r/Fi
ller/
Vf
/Wf
%E
rode
ntm
ater
ial
Ero
dent
size
(μm
)A
ngle
(deg
)V
eloc
ity(m
/s)
Yea
rof
Ref
eren
ce
publ
icat
ion
PEE
K/P
EK
K66
–68
%(w
f)
unid
irec
tiona
lGF,
70–7
1%
(wf)
Silic
aSa
nd15
0–25
015
,30,
60,9
025
,37,
50,6
020
09[5
9]
unid
irec
tiona
lCF
PPS
0–40
%(w
f)
shor
tgla
ssfi
ber
Silic
aSa
nd15
0–20
015
,30,
60,9
020
,40,
6620
09[6
0]
EP
5%
,10
%,1
5%
(wf)
shor
tfla
kes
obta
ined
from
fish
scal
eD
rysi
lica
sand
300,
500,
800
30,6
0,90
32,4
4,58
2009
[61]
EP
30%
,405
,50
%(w
f)
GF,
10%
(wf)
TiO
2Si
lica
sand
400,
500,
600
30,6
0,90
45,6
5,85
2009
[62]
PET
50%
(wf)
GF,
10%
,20
%(w
f)
fly
ash,
Al 2
O3,S
iCD
rysi
lica
sand
300,
500,
800
40,6
0,90
32,4
5,58
2010
[63]
EP
10%
,20
%(w
f)
red
mud
,50
%(w
f)
bam
boo
and
glas
sfi
ber
Dry
silic
asa
nd30
0,45
0,60
030
,60,
9043
,54,
6520
10[6
4]
UH
MW
PE,E
PA
ram
idfi
ber
Silic
asa
nd30
,50,
60,8
030
,90
Con
stan
t20
10[6
5]
EP
Uni
dire
ctio
nalg
lass
fibe
rSi
lica
sand
150
±15
30,6
0,90
–20
11[6
6]
EP
15%
(wf)
bori
cac
idfi
ller,
Gla
ssfi
ber
Al 2
O3
200,
400
30,6
0,90
23,3
4,53
2011
[67]
EP
15%
(wf)
Silic
onox
ide
fille
r,G
lass
fibe
rSi
O2
200,
400
30,6
0,90
23,3
4,53
2011
[68]
EP
50%
,75
%(w
f)
jute
fibe
r,50
%,2
5%
(wf)
glas
sfi
ber
Silic
asa
nd15
0–25
030
,45,
60,9
048
,70,
8220
11[6
9]
Vin
yles
ter
20%
,30
%,4
0%
,50
%(w
f)
carb
onfi
ber
Dry
silic
asa
nd25
0,35
5,42
0,60
030
,45,
60,9
043
,54,
65,7
620
11[7
0]
occur. From medium to high velocities, a power law [74]can describe the relationship between wear rate and impactvelocity:
−dm
dt= kνn (1)
Where m is the mass of the worn specimen, t is the dura-tion of the process, k is an empirical constant, n is a velocityexponent. The characteristics of the erodent and that of thetarget material determine the value of the exponent ‘n’. Ithas been stated that ‘n’ varies in the range of 2–3 for poly-meric materials behaving in a ductile manner, while forpolymer composites behaving in brittle fashion the value of‘n’ is in the range of 3–5 [3, 33]. Patnaik et al. [55] haveextensively studied reported on the effect of impact velocityin glass polyester composites with and without a variety ofhard ceramic fillers.
4.2 Effect of Angle of Impingement
The erosion wear rates of the composites are found to bedependent on the impingement angle. The findings of thisresearch further suggest that, this dependency is also influ-enced by the nature of the filler material. In fact, the angle ofimpact determines the relative magnitude of the two compo-nents of the impact velocity namely, the component normalto the surface and parallel to the surface. The normal com-ponent will determine how long the impact will last (i.e.contact time) and the load. The product of this contact timeand the tangential (parallel) velocity component determinesthe amount of sliding that takes place. The tangential veloc-ity component also provides a shear loading to the surface,which is in addition to the normal load that the normalvelocity component causes. Hence, as this angle changes theamount of sliding that takes place also changes the natureand magnitude of the stress system. Both of these aspectsinfluence the way a composite wears out. Studies made bya number of recent investigators imply that composites withfillers of different type and content would exhibit differentangular dependency.
4.3 Effect of Erodent Type, Size and Shape
The type and physical characteristics of the erodent materialplay a key role in the erosion problem. Variations in ero-dent particle size and shape can cause fundamental changesin the erosion response of polymer composites [75]. Transi-tions in wear mechanisms can often be attributed to a changein the shape, hardness or size of the erodents [74, 76]. If theeroding particles are blunt or spherical then plastic defor-mation is favored, if the particles are sharp then cutting andbrittle fragmentation are more likely. A blunt particle has amostly curved surface approximating to a spherical shape
Silicon (2014) 6:5–20 11
while a sharp particle consists of flat areas joined by cor-ners with small radii which are critical to the process ofwear. It is assumed that the ER is independent of particlesize above a critical value [26, 76, 77]. This critical value isobserved between 100 and 200 μm, however, it is dependenton the exposure conditions and the particle target interac-tion [75, 77]. Up to this critical value, experimental resultsshowed that with increasing size of the erodent also the ERincreases.
A trend of reverse nature is observed after this criticalsize, because of the increased possibility of particle colli-sions as the erodent size increases. Another possible reasonis that less number of particles reached the sample per unitweight of erodent when the particles impacting the samplehad larger sizes. In that case the larger particles became lesseffective and that results in a lower ER [78].
4.4 Effect of Erodent Hardness and Fracture Toughness
The erodent fracture toughness may influence the erosiveprocedure if fragmentation of the erodent occurs duringimpact. When a particle breaks into several fragments, theinitial energy and therefore the stresses on the surface aredistributed over a larger area. Additionally, these cleavage-processes reduce the part of the energy getting into thematerial. From energetic view, the fracture of the parti-cle has a wear-reducing effect. However, if the erodentfragments produced have sharper edges, compared to theoriginal particles, then wear may also increase [79]. Theeffect of erodent hardness depends mainly on the particularmode of erosive wear taking place, e.g., ductile or brittle. Inthe brittle mode the effect of particle hardness is much morepronounced than in the ductile mode. It is usually believedthat hard particles cause higher wear rates than soft ones, butit is impossible to isolate hardness completely from otherfeatures of the particle (e.g., shape). Even if the particle ishard, but relatively blunt, then it is unlikely to cause severeerosive wear [74]. With respect to the size and type of theerodent material, two trends may hold for harder and/ormore brittle material. The erosive wear increases the higherthe hardness of the erodent and the larger the erosive par-ticle size are (until a level of saturation is reached in bothcases) [75]. In ductile polymers, however, the situation maybe quite different. Due to the relatively low hardness no pro-nounced effects of changes in the hardness of the usuallymuch harder erodent materials should be expected [75].
4.5 Effect of Erodent Feed Rate
The particle feed rate (i.e., the mass of impacting mate-rial per unit area and time) is another controlling parameterof the erosive wear rate. Theoretically, the ER should beindependent of the flux of particles striking a target since
(it is assumed that) all the particles hit the target with thesame velocity and angle of impact. In practice, however,significant effects of particle flux on measured ER wereobserved [74, 80–82]. It has been reported that erosivewear rate was proportional to the feed rate up to a cer-tain threshold value. This limit is believed to be the resultof interference between rebounding- and arriving-particles[74]. This effect is rationalized in terms of a first-orderparticle collision model where the collisions removed theincident particles from the erosion process [82]. This colli-sion effect may be significant, even for relatively low valuesof feed. The limiting particle feed rate is highly variable,ranging from as low as 100 kg/m2s for elastomers to as highas 10,000 kg/m2s for erosion against metals by large and fastparticles. The wear rates decrease marginally when the lim-iting feed rate is exceeded [74, 80]. However, although theeffect of erodent feed rate has been mainly attributed to theabove mentioned interaction, there may also be other pos-sible mechanisms that can take place and affect the erosivewear.
4.6 Effect of Erodent Temperature
Erosion of the target material occurs when a solid particleimpinges on it with some significant velocity. In other wordsthe kinetic energy of the impacting solid causes the materialloss which is termed as erosion wear. But when these solidparticles are at an elevated temperature (higher than the tar-get surface temperature), they also dissipate a part of theirthermal energy in addition to the kinetic energy to the tar-get material causing greater damage to the surface. Biswaset al. [83] have reported extensively on the effect of ero-dent temperature on the erosion wear of a variety of epoxybased composites. The findings of these studies reveal thatthe erosion wear rates of polymer composites increase withthe increase in the erodent temperature. The existing dataon the effect of temperature on solid particle erosion arerationalized in a qualitative fashion by means of the local-ization model for erosion. It is shown that the temperaturedependence of the flow stress of the eroding material andthe extent of variation in the strain hardening exponent withtemperature largely determine the influence of temperatureon erosion rate.
4.7 The Effect of the Fiber Material
More than 90 % of the composites discussed are reinforcedwith either glass fibers (GF) or carbon fibers (CF). Fewerstudies can be found on aramid fibers (AF) or other typeof reinforcement. In most studies the fiber reinforced com-posites are compared with the unreinforced matrices. Thetrend is that the addition of the fibers, which are mostcommonly brittle in nature, leads to deterioration of the
12 Silicon (2014) 6:5–20
erosion resistance of the matrix. This holds especially whena thermoplastic matrix is used.
4.8 The Effect of the Fiber Content
Although this parameter cannot be isolated from otherparameters like the fiber brittleness, the general conclusionis that the higher the fiber content, the lower the erosionresistance. Several studies have tried to apply modified ruleof mixtures in order to correlate the ER with the fibercontent. These studies have been reviewed in details in [84].
4.9 The Effect of the Reinforcement Type(i.e. Length, Diameter, Weave Style etc.)
Most studies focus on long continuous fibers. Short-chopped fibers come next, while also woven type of rein-forcement has been of interest (plain weave). Generally,short fiber reinforced composites show better resistance toerosion compared to long-continuous fiber reinforced ones.Ductile fiber reinforced (even self-reinforced) polymers arealso termed as self-healing polymers. Self-healing is receiv-ing an increasing amount of interest worldwide as a methodto address damage in materials. In particular, for advancedhigh-performance fiber-reinforced polymer (FRP) compos-ite materials, self-healing offers an alternative to employingconservative damage-tolerant designs and a mechanism forameliorating inaccessible and invidious internal damagewithin a structure.
4.10 The Effect of the Relative Fiber OrientationDuring Erosion
These studies focus mainly on unidirectional composites.Here the reinforcement can be parallel or perpendicular tothe impingement direction, or even at 45◦. The results foundin the literature show different trends depending on vari-ous parameters like the angle of impingement, the fibersductility, the fiber/matrix adhesion, the fiber content, etc.Therefore, it is a parameter that cannot be considered iso-lated. It is true that the fibers can be at any orientationto the erodent. Or in other words, the erodent stream maystrike the fibers at any angle with respect to the longitu-dinal direction of the fibers. But in case of experimentalworks, there is always a restriction to the number of lev-els that can be taken for a particular factor. Most of thereported investigations are therefore confined to 0◦, 45◦ and90◦ of fiber orientations to the erodent. But the investiga-tion can be extended to any other values as well. Brandtet al. [85] proposed that fiber type, form, and orientation(fiber architecture) comprise the main considerations whenchoosing reinforcements. In a part that will be carrying lit-tle or no structure loads, chopped or continuous strand mat
with random fiber orientation is sufficient. However, in apart that will see primary or secondary structural loads, fiberorientation is critical and departures from the optimum canresult in drastic property reduction. Fiber architecture canbe tailored for specific requirements, with parallel longitudi-nal (0◦ strands carrying tension loads, circumferential (90◦)strands providing compression and impact strength, andhelical (commonly ±33◦ or ±45◦) strands handling torquestresses. This design principle is comparable to the waythat civil engineers use steel-reinforcing bar in a concretestructure.
5 Process Variables in Erosion Wear Modeling
Several erosion models/correlations were developed bymany researchers to provide a quick answer to design engi-neers in the absence of a comprehensive practical approachfor erosion prediction. The theoretical model developed byRabinowicz [86] was used to calculate the volume of mate-rial removed from the target surface due to impact of solidparticles entrained in a liquid jet. The results indicated thatthe sand particle trajectories appeared to be governed by thesecondary flows and that there was no simple liquid velocityprofile that can be used to calculate the particle trajectoriesin order to make an accurate prediction of the location of thepoint of maximum wear. One of the early erosion predictioncorrelations was developed by Finnie [87] expressing therate of erosion in terms of particle mass and impact veloc-ity. In that correlation, the rate of erosion was proportionalto the impact velocity squared. In a recent study, Nesic [88]found that Finnie’s model over-predicts the erosion rate andpresented another formula for the erosion rate in terms of acritical velocity rather than the impact velocity. The erosionmodel suggested by Bitter [89, 90] assumed that the ero-sion occurred in two main mechanisms: the first was causedby repeated deformation during collisions that eventuallyresulted in a piece of material breaking loose while the sec-ond was caused by the cutting action of the free-movingparticles. Comparisons between the obtained correlationsand the test results showed a good agreement. It was con-cluded that cutting wear prevails in places where the impactangles are small (such as in risers and straight pipes) and it issufficient to use hard material in such places to reduce ero-sion. Other erosion models were suggested by Laitone [91],Salama and Venkatesh [92], Bourgoyne [93], Chase et al.[94], McLaury [95], Svedeman and Arnold [96], and Jordan[97]. Recently, Shirazi and McLaury [98] presented a modelfor predicting multiphase erosion in elbows. The modelwas developed based on extensive empirical informationand it accounts for the physical variables affecting erosion,including fluid properties, sand production rate and size andthe fluid-stream composition. An important feature of this
Silicon (2014) 6:5–20 13
model was the use of the characteristic impact velocity ofthe particles.
In most erosion processes, target material removal typ-ically occurs as the result of a large number of impactsof irregular angular particles, usually carried in pressur-ized fluid streams. The fundamental mechanisms of materialremoval, however, are more easily understood by analy-sis of the impact of single particles of a known geome-try. Such fundamental studies can then be used to guidedevelopment of erosion theories involving particle streams,in which a surface is impacted repeatedly. Single parti-cle impact studies can also reveal the rebound kinematicsof particles, which are very important for models whichtake into account the change in erosive potential due tocollisions between incident and rebounding particles [99,100]. A number of recent articles contain investigations ofthe rebound kinematics of spherical/angular particles [101–105]. These are concerned with the identification and mod-eling of mechanisms of ductile target material removal dueto the impact of single hard, angular particles. Finnie’s anal-ysis [87] of the cutting action of a single particle launchedagainst a ductile target was the first such model capable ofpredicting material removal rate. In this model, the parti-cles were assumed to be rigid and impact a target whichreached a constant flow pressure immediately upon impact.Under the assumption that the particle did not rotate duringthe impact process, the particle was subjected to a resistingforce vector of constant direction, and Finnie was able tosolve for the trajectory of the particle in closed form as it cutthe surface, and thus predict the size of the impact crater. Anextension of Finnie’s work, the rigid-plastic theory devel-oped by Hutchings et al. [106–108], predicted the collisionkinematics and crater dimensions for single impacts ofsquare and spherical particles on ductile targets. The theorypredicted the kinematics of the particle as it ploughed or cutthrough the target, under the assumption that the instanta-neous resisting force could be calculated by multiplying aconstant plastic flow pressure (i.e., the dynamic hardness)by the instantaneous contact area. In contrast to the con-stant direction force vector assumed in Finnie’s work and inHutchings’ analysis, the particle was free to rotate, and theresisting force vector could thus vary in both direction andmagnitude. By examining the single impacts of the squareand spherical particles, Hutchings identified two fundamen-tal mechanisms of cutting erosion, and a ploughing erosionmechanism, depending on both the particle shape and itsorientation at the moment of impact [107]. In general, com-parisons of experimentally measured crater volume, energyloss, and particle kinematics yielded acceptable agreement.For impacts involving spherical particles, the rigid plas-tic theory was later improved by Ricker by and Macmillan[109, 110], to include a more accurate calculation of con-tact area. Sundararajan et al. [111–113] also used a similar
theory to model ductile erosion, and investigated the effectof material pile-up at the edge of the crater on the reboundkinematics of the spherical particles, and the size of theresulting plastic zone below the impact. In the case of therigid-plastic impact of single angular particles, however,very little literature exists. A rigid-plastic theory developedby Papini et al. [114–117] generalized Hutchings’ [80, 118]rigid-plastic theory for square particles, so that particlesof any shape impacting targets of arbitrary dynamic hard-ness and dynamic friction coefficient could be analyzed.The specific case of two dimensional particles having rhom-boidal shape (i.e., ‘diamond shaped’) of varying angularitywas studied in detail by constructing a computer programcapable of describing the trajectory of the particles as theyformed impact craters, so that their size and shape could bepredicted [116]. Dimensionless parameters were identifiedso that the results of a parametric study could be presentedin a generally applicable form, fundamental erosion mecha-nisms were predicted, and it was postulated that for a givenangle of attack, there was an optimum particle shape for themost efficient material removal [117].
In order to develop a mathematical model, it is importantto understand the mechanism responsible for solid-particleerosion of composite materials. For a composite material,its surface damage by solid-particle erosion depends onmany factors, including the impact velocity, particle sizeand shape of the erodent, mechanical properties of boththe target material and the erodent, and the volume frac-tion, size and properties of the reinforcing phase as wellas the bonding between the matrix and the reinforcingphase. The synergism of these factors makes it difficult toexperimentally investigate the erosion mechanism for com-posite materials. Fortunately, computer simulation providesan effective and economic approach for such investiga-tion. Computer models proposed to simulate wear processmay be classified into two groups: macro-scale models andatomic-scale models. The macro-scale models were pro-posed based on various assumptions or theories such as thecutting mechanism [87] and the platelet mechanism [119].The cutting mechanism is based on the assumption that indi-vidual erodent particle impinges a target surface, cutting outa swathe of the material. However, this mechanism is onlysuitable for ductile materials. Even for ductile materials,SEM observation of eroded surfaces has shown that erosionprocesses of metals involve extrusion, forging, and fracture,and that micro-cutting does not often occur [120]. Regard-ing the platelet mechanism, plastic deformation and workhardening prior to fracture are taken into account and thismakes it closer to reality. However, this mechanism is alsoonly suitable for ductile materials.
Recently, Patnaik, Satapathy, Mahapatra, and Dash [57,121] developed a theoretical model to estimate the ero-sion wear rate of polymer composites under multiple impact
14 Silicon (2014) 6:5–20
condition. The model is based on the assumption that thekinetic energy of the impinging particles is utilized to causemicro-indentation in the composite material and the mate-rial loss is a measure of the indentation. The erosion is theresult of cumulative damage of such non-interacting, singleparticle impacts. The model further assumes the erodent par-ticles to be rigid, spherical bodies of diameter equal to theaverage grit size. It considers the ductile mode of erosionand assumes the volume of material lost in a single impact isless than the volume of indentation. It further considers thatthe hardness alone is unable to provide sufficient correla-tion with erosion rate, largely because it determines only thevolume displaced by each impact and not really the volumeeroded. Thus a parameter which will reflect the efficiencywith which the volume that is displaced is removed shouldbe combined with hardness to obtain a better correlation.The ‘erosion efficiency’ is obviously one such parameter. Incase of a stream of particles impacting a surface at any angleα, the erosion efficiency defined by Patnaik et al. [57, 121]is given as:
η = 2ErHν
ρV 2 Sin2 α(2)
and according to the model proposed by them, the non-dimensional erosion wear rate of a composite material isgiven by:
Erth = ρcηV 2 sin2 α
2Hv
(3)
where α is the angle of impingement (degree), V is theimpact velocity (m/s), Hv is the hardness (N/m2), ρc is thedensity of composite (kg/m3), ρ is the density of erodent(kg/m3), η is the erosion efficiency, Er is the actual erosionwear rate (kg/kg), and Erth is the theoretical erosion wearrate (kg/kg).
The magnitude of η can be used to characterize thenature and mechanism of erosion. For example, ideal micro-ploughing involving just the displacement of the materialfrom the crater without any fracture (and hence no erosion)will results in η = 0. In contrast, if the material removalis by ideal micro-cutting, η = 1.0 or 100 %. If erosionoccurs by lip or platelet formation and their fracture byrepeated impact, as is usually the case in ductile materi-als, the magnitude of η will be very low, i.e., η ≤ 100 %.In the case of brittle materials, erosion occurs usually byspalling and removal of large chunks of materials resultingfrom the interlinking of lateral or radial cracks and thus η
can be expected to be even greater than 100 %. Patnaik et al.[54–125] conducted erosion trials on a number of polyester-based composites and demonstrated that if supported byan appropriate magnitude of erosion efficiency, the abovemodel performed well for polyester matrix composites fornormal as well as oblique impacts.
But the model proposed by Patnaik et al. [57] assumesthat both the erodent material and the target material are atsame temperature and therefore there is no exchange of anythermal energy between them during the impact. This maybe true for a room temperature erosion situation, but whenthe erodent is at an elevated temperature, as in the case ofhot air carrying pulverised coal powders in a pipe, therewill be dissipation of the kinetic energy as well as the ther-mal energy from the erodent body to the target. Researchon erosion of composite materials by high temperature ero-dent particles has been rare but very recently Biswas andSatapathy [126] have proposed another model which takesinto account this approach of energy dissipation. Besides,while all previous models have been developed assumingthe shape of erodent to be spherical, in the real situation, theerodent particles are actually irregular shaped bodies hav-ing sharp edges. Considering them to be square pyramidalshaped bodies is a more realistic assumption as comparedto assuming them simply spherical. The model proposed byBiswas and Satapathy [126] addresses to this shortcomingas well. It assumes the erodent particles to be rigid, squarepyramidal shaped bodies of height and base length equal tothe average grit size. As already mentioned, it also assumesthat the loss in both kinetic as well as thermal energy of theimpinging particles is utilized to cause micro-indentation inthe composite material and the material loss is a measureof the indentation. The erosion is the result of cumulativedamage of such non-interacting, single particle impacts. Themodel is developed with the simplified approach of energyconservation which equals the loss in erodent kinetic energyand thermal energy during impact with the work done increating the indentation. According to this model, the non-dimensional erosion rate, defined as the composite masslost per unit time due to erosion divided by the mass of theerodent causing the loss, is expressed as
Er = ηρC
3H
[U2sin2α + 2S (θ − θ0)
](4)
where: U: impact velocity (m/s),θ : erodent temperature (◦C)and, θ0: room temperature (◦C). The mathematical expres-sion in Eq. (4) can be used for predictive purpose to makean approximate assessment of the erosion damage from thecomposite surface. Biswas and Satapathy [126] conductederosion trials on a set of particulate filled epoxy compositesand demonstrated that if supported by an appropriate mag-nitude of erosion efficiency, the model performed well fornormal as well as oblique impacts.
6 Erosion Modeling Based on Averaging Rules
When designing with composites it is important to knowwhat will be the overall ER in the combined multiphase
Silicon (2014) 6:5–20 15
material if the ERs of the individual constituent are given.As an answer to this, the linear (LROM) and inverse (IROM)rule of mixture have been proposed for the prediction of theER.
Equations (5) and (6) describe the LROM and the IROM,respectively:
ERc = wf × ERf + wm × ERm (5)
1
ERc
= wf
ERf
+ wm
ERm
(6)
where subscripts c, f and m mean composite, fibre andmatrix respectively, whereas ER and w denote the erosionrate and the weight fraction of the related material.
These two rules of mixture have also been proposed tomodel the wear of UD fibre reinforced composite materi-als [127]. The LROM and IROM were first evaluated fora multiphase AL-Si alloy [128]. The same rules of mixturewere adopted for a glass-fibre reinforced epoxy composite[129]. The key aspect in the problem of the ‘averaging law,’is the size of the impact site in comparison to the size of themicrostructural phase [128]. This work enlightened and dis-cussed the different cases arising from the microstructure ofthe composite and from the behaviour of each constituent.Generally an increase of the fibre/filler content leads to anincrease to the ER. This is due to the fact that usually theerosion resistance of the fibres is lower than that of thematrix. A further reason is the quality of the bonding of thereinforcement with the matrix [34, 130–132]. In case of par-ticulate composites the interface between matrix and filleris not only weak but may also promote subsurface crackpropagation. This will accelerate further the ER. Such phe-nomena have been reported in case of rubbers reinforcedwith fillers [133, 134].
Generally the results showed that the linear rule of mix-ture and the inverse rule of mixture provide good bounds forthe experimental ER. The inverse rule of mixture deliversgenerally better results. The modified rules of mixture pro-posed for the case of abrasion do not hold for the erosivewear. The applicability of the LROM in some of the exper-imental results [23] verified the already existing remark[128], that although generally the IROM predicts better theER of multiphase systems, when the two constituents arecontinuous and linear aligned along the incident erodentparticle beam direction, the LROM approach works well.
7 Implementation of Design-of-Experiments for WearAnalysis
Wear processes in composite materials are complex phe-nomena involving a number of operating variables and
it is essential to understand how the wear characteristicsare affected by different operating conditions. Althoughmany researchers have reported on the properties, perfor-mance and wear characteristics of materials, the signif-icance of different process parameters and their relativeinfluence on wear rate has not adequately been studiedyet. Selecting the correct operating conditions is alwaysa major concern as traditional experimental design wouldrequire many experimental runs to achieve a satisfactoryresult. In any process, the desired testing parameters aredetermined either based on experience or by use of ahandbook. It, however, does not provide optimal testingparameters for a particular situation. Thus, several math-ematical models based on statistical regression techniqueshave been constructed to select the proper testing con-ditions [135–140]. Design-of-Experiments (DOE) is aneffective method for conducting experiments to improvethe performance output by finding optimum test condi-tions. This also provides a better understanding of physicalmechanisms involved in a complex phenomenon like a wearprocess.
Statistical methods have commonly been used for anal-ysis, prediction and/or optimization of a number of engi-neering processes. These methods enable the user to defineand study the effect of every single condition possiblein an experiment where numerous factors are involved.Wear processes in composites are such complex phenom-ena involving a number of operating variables and it isessential to understand how the wear characteristics ofthe composites are affected by different operating condi-tions. Selecting the proper operating conditions is alwaysa major concern as traditional experiment design wouldrequire many experimental runs to achieve satisfactoryresult. Usually in an experimental trial, the number ofruns required for a full factorial design increases geo-metrically whereas fractional factorial design significantlyreduces the number of runs and saves time. This methodis popular because of its simplicity, but this very simplic-ity has led to unreliable results and inadequate conclusions.The fractional design might not contain the best designpoint. Moreover, the traditional multi-factorial experimen-tal design is the change-one-factor-at-a-time method. Underthis method, only one factor is varied at a time, whileall the other factors are kept fixed at a specific set ofconditions.
To overcome these problems, Taguchi and Konishi[141] advocated the use of orthogonal arrays and Taguchi[142] devised a new experimental design that appliedsignal-to-noise ratio with orthogonal arrays to the robustdesign of products and processes. In this procedure, theeffect of a factor is measured by average results andtherefore, the experimental results can be reproducible.Phadke [143], Wu and Moore [144] and others [127,
16 Silicon (2014) 6:5–20
145] have subsequently applied the Taguchi method todesign the products and process parameters. This inex-pensive and easy-to-operate experimental strategy basedon Taguchi’s parameter design has been adopted to studyeffect of various parameters and their interactions in anumber of engineering processes. It has been applied forparametric appraisal in wire electrical discharge machin-ing (WEDM) process, drilling of metal matrix compositesand for optimization of cutting parameters in turning oper-ations [55, 146–152]. Reports are also available on suc-cessful implementation of Taguchi experimental design insolid particle erosion of multiphase composite materialssuch as glass-polyester-alumina, glass-polyester-SiC, glass-polyester-fly ash, glass-epoxy-red mud composites etc.[124, 126, 153]. These studies suggest that use of Taguchimethod leads to the identification of significant controlfactors responsible for material loss due to wear. It alsohelps to determine optimal factor settings for minimumwear.
8 Use of Neural Computation for Wear Prediction
Erosion wear process is considered as a non-linear problemwith respect to its variables: either materials or operatingconditions. To obtain minimum wear rate, combinationsof operating parameters have to be planned. Therefore arobust methodology is needed to study these interrelatedeffects. In this work, a statistical method, responding tothe constraints, is implemented to correlate the operatingparameters. This methodology is based on ANN, whichis a technique that involves database training to predictinput–output evolutions.
The use of neural networks represents a new methodol-ogy in many different applications. It is a promising field ofresearch in predicting experimental trends and has becomeincreasingly popular in the last few years as they can oftensolve problems much faster compared to other approacheswith the additional ability to learn. Generally, a neural net-work means a network of many simple processors (“units”)operating in parallel, each possibly having a small amountof local memory. The units are connected by communica-tion channels (“connections”) which usually carry numeric(as opposed to symbolic) data, encoded by one of variousways [154]. One of the best known examples of a biolog-ical neural network is the human brain. It has the mostcomplex and powerful structure which, by learning andtraining, controls human behaviour towards responding toany problem encountered in every-day life. As for the arti-ficial neural networks (ANN), they have been developed toemulate this biological network for the purpose of learn-ing the solution to a physical problem from a given setof examples.
Artificial neural networks are comparatively new mod-elling techniques, which can be used to solve problemsthat are difficult for conventional computers or humanbeings. The ANNs have been applied to model complicatedprocesses in many engineering fields, such as aerospace,automotive, electronics, manufacturing, robotics, telecom-munications etc. For predictive purposes, an ANN approachhas also been introduced recently into the field of wear ofpolymers and composites by Velten et al. [155] and Zhanget al. [156].
As mentioned earlier, an ANN is a computational sys-tem that simulates the microstructure (neurons) of thebiological nervous system. It has shown remarkable per-formance when used to model complex linear and non-linear relationships. The ANNs offer a fundamentally dif-ferent approach to material modelling and material pro-cessing control techniques than the statistical methods andrecently the interest in the ANN modelling in the fieldsof materials science and physical metallurgy has increased[157–161]. The multi-layered neural network is the mostwidely applied neural network, which has also beenutilized in research works related to polymer composites[162].
9 Conclusions
The present study was aimed at reviewing the role of pro-cess variables on the solid the particle erosion responseof polymer and their composites focusing on the dominat-ing mechanism, the most discussed influencing parametersand the different trends observed in the literature. The lit-erature survey presented in this paper reveals the growingresearch in the field of erosion of composites particularlyafter year 2000. But although a great amount of workhas already been devoted to this topic many questionsare still open. A comprehensive and systematic investi-gation of erosion in polymer composites has not beenperformed yet. Studies made on the erosive wear of com-posites refer more on fibre-reinforced polymer (FRP) andless on filler-reinforced-systems. The effect of fillers isconsidered more as modification of the matrix and lessas reinforcement, possibly because of the low percentageof fillers. As a result, the effect of particulate fillers onerosion characteristics of hybrid composites has receivedmuch less research attention. There is no clear understand-ing of the mechanism of erosion and how the propertiesof the constituents and the interface affect the erosionbehaviour of these composites. Extensive research is there-fore needed to develop various methods and theoreticalmodels for predicting erosion behaviour and its dependenceon the proportion of the components and the compositemicrostructure.
Silicon (2014) 6:5–20 17
References
1. Finnie I (1996) Some reflections on the past and future oferosion. Wear 186(187):1–10
2. Meng HC, Ludema KC (1995) Solid particle erosion resistanceof ductile wrought super alloys and their weld overlay coatings.ibid 443:181–183
3. Pool KV, Dharan CKH, Finnie I (1986) Erosion wear of com-posite materials. Wear 107:1–12
4. Aglan HA, Chenock Jr TA (1993) Erosion damage features ofpolyimide thermoset composites. SAMPE Q 24:41–47
5. Rao PV (1995) Characterization of optical and surface param-eters during particle impact damage. ASME/Fluids Eng Publ23:87–96
6. Tennyson RC (1991) LDEF mission upyear: composites in space.Adv Mater Process 5:33–36
7. Kulkarani SM (2001) Influence of matrix modification on thesolid particle erosion of glass/epoxy composites. Polym Compos9:25–30
8. Tilly GP (1973) A two stage mechanisms of ductile erosion.Wear 23:87–96
9. Ruff AW, Wiederhorn SM (1979) Erosion by solid particleimpact. In: Preece CM (ed) Treatise on materials science andtechnology, vol 16. Academic, New York, pp 69–125
10. Humphrey JAC (1990) Fundamentals of fluid motion inerosion by solid particle impact. Int J Heat Fluid Flow11(3):170–195
11. Hollaway L (1994) Handbook of polymer composites for engi-neers. Woodhead Publishing Ltd., Cambridge, p 1
12. Pritchard G (1999) Reinforced plastics durability. WoodheadPublishing Ltd., Cambridge, p 1
13. Mason JS, Smith BV (1972) The erosion of bends by pneu-matically conveyed suspensions of abrasive particles. PowderTechnol 6:323–335
14. Tsai W, Humphrey JAC, Cornet I, Levy AV (1981) Experimentalmeasurement of accelerated erosion in a slurry pot tester. Wear68:289–303
15. Hojo H, Tsuda K, Yabu T (1986) Erosion damage of polymericmaterial by slurry. Wear 112:17–28
16. Kumar R, Verma AP, Lal GK (1983) Nozzle wear during theflow of a gas–particle mixture. Wear 91:33–43
17. Crowley MS (1969) Influence of particle size on erosion resis-tance of refractory concretes. Am Ceram Soc Bull 48:707–710
18. Wright IG (1979) Proceedings corrosion/erosion of coal conver-sion systems materials. Conference, Berkeley, pp 103–138
19. Jansson SA (1982) Proceedings corrosion-erosion-wear of mate-rials in emerging fossil energy systems. Berkeley, pp 548–560
20. Neilson JH, Gilchrist A (1968) An experimental investigationinto aspects of erosion in rocket motor tail nozzles. Wear 11:123–143
21. Hibbert WA (1965) Helicopter trials over sand and sea. J RoyAeronaut Soc 69:769–776
22. Miyazaki N, Hamaom T (1994) Solid particle erosion of thermo-plastic resins reinforced by short fibres. J Comp Mater 28:871–883
23. Barkoula NM, Karger-Kocsis J (2002) Effect of fibre content andrelative fibre orientation on the solid particle erosion of gf/ppcomposites. Wear 252:80–87
24. Tilly GP (1969) Erosion caused by airborne particles. Wear14(1):63–79
25. Zahavi J, Nadiv S, Schmitt Jr GF (1981) Indirect damage incomposite materials due to raindrop impact. Wear 72(3):305–313
26. Tilly GP, Sage W (1970) The interaction of particle and materialbehaviour in erosion processes. Wear 16(6):447–465
27. Tsiang TH (1989) Sand erosion of fiber composites: testing andevaluation. In: Chamis CC (ed) Test methods for design allow-able for fibrous composites, vol 2. American Society for Testingand Materials, (ASTM) STP I003, Philadelphia, pp 55–74
29. Latif A, Ahmed A (1987) Geostatistical estimation of reserves inthe abu-tartur phosphate deposits western desert, Egypt. Mastersthesis, King Fahd University of Petroleum and Minerals
30. Karasek KR, Goretta KC, Helberg DA, Routbort JL (1992)Erosion in bismaleimide polymers and bismaleimide polymercomposites. J Mater Sci Lett 11:1143–1144
31. Tilly GP (1969) Sand erosion of metals and plastics: a briefreview. Wear 14(4):241–248
32. Brandstadter A, Goretta KC, Routbort JL, Groppi DP, KarasekKR (1991) Solid particle erosion of bismaleimide polymers.Wear 147:155–164
33. Roy M, Vishwanathan B, Sundararajan G (1994) The solidparticle erosion of polymer matrix composites. Wear 17:149–161
34. Miyazaki N, Hamao T (1996) Effect of interfacial strength onerosion behaviour of frps. J Comp Mater 30:35–50
43. Srivastava VK (2006) Effects of wheat starch on erosive wear ofe-glass fiber reinforced epoxy resin composite materials. MaterSci Eng A 435–436:282–287
44. Sınmazcelik T, Taskıran I (2007) Erosive wear behaviour ofpolyphenylenesulphide (pps) composites. Mater Des 28:471–2477
45. Yang N, Nayeb-Hashemi H (2007) The effect of solid particleerosion on the mechanical properties and fatigue life of fiber-reinforced composites. J Compos Mater 41:559–574
46. Rattan R, Bijwe J (2007) Influence of impingement angle onsolid particle erosion of carbon fabric reinforced polyetherimidecomposite. Wear 262:568–574
47. George K (2002) Experimental investigation of composite mate-rial, erosion characteristics under conditions encountered inturbofan engines. Doctor of Philosophy. M.S. University ofCincinnati, Cincinnati
18 Silicon (2014) 6:5–20
48. Sarı N, Sınmazcelik T (2007) Erosive wear behaviour of carbonfibre/polyetherimide composites under low particle speed. MaterDes 28:351–355
49. Harsha AP, Thakre AA (2007) Investigation on solid particleerosion behaviour of polyetherimide and its composites. Wear262:807–818
50. Sinmazcelik T, Fidan S, Gunay V (2008) Residual mechani-cal properties of carbon/polyphenylenesulphide composites aftersolid particle erosion. Mater Des 29:1419–1426
51. Arjula S, Harsha AP, Ghosh MK (2008) Erosive wear of uni-directional carbon fibre reinforced polyetherimide composite.Mater Lett 62:3246–3249
52. Yang NH, Nayeb-Hashemi H, Vaziri A (2008) Non-destructiveevaluation of erosion damage on e-glass/epoxy composites.Compos Part A 39:56–66
53. Harsha AP, Jha SK (2008) Erosive wear studies of epoxy-basedcomposites at normal incidence. Wear 265:1129–1135
54. Patnaik A, Satapathy A, Mahapatra SS, Dash RR (2008) Ataguchi approach for investigation of erosion of glass fiber-polyester composites. J Reinf Plast Compos 27(8):871–888
55. Patnaik A, Satapathy A, Mahapatra SS, Dash RR (2008) Para-metric optimization of erosion wear of polyester-gf-aluminahybrid composites using taguchi method. J Reinf Plast Compos27(10):1039–1058
56. Patnaik A, Satapathy A, Mahapatra SS, Dash RR (2008) Imple-mentation of taguchi design for erosion of fiber reinforcedpolyester composite systems with SiC filler. J Reinf Plast Com-pos 27(10):1093–1111
57. Patnaik A, Satapathy A, Mahapatra SS, Dash RR (2008) A mod-eling approach for prediction of erosion behaviour of glass fiber-polyester composites. J Polym Res 15(2):147–160
58. Kim A, Kim L (2009) Solid particle erosion of cfrp compositewith different laminate orientations. Wear 262:1922–1926
60. Suresh A, Harsha AP, Ghosh MK (2009) Solid particle erosionstudies on polyphenylene sulfide composites and prediction onerosion data using artificial neural networks. Wear 266:184–193
61. Satapathy A, Patnaik A, Pradhan MK (2009) A study on pro-cessing, characterization and erosion behaviour of fish (labeo-rohita) scaled filled epoxy matrix composite. Mater Des30:2359–2371
62. Biswas S, Ray S, Satapathy A, Patnaik A (2009) Erosion wearbehaviour of tio2 filled glass fiber reinforced epoxy composite.Material Sci An Indian J 5(3):1–9
63. Patnaik A, Satapathy A, Biswas S (2010) Effect of particulatefillers on erosion wear of glass polyester composites: a compar-ative study using taguchi approach. Malaysian Polym J 5(2):49–68
64. Biswas S, Satapathy A (2010) A comparative study on erosioncharacteristics of red mud filled bamboo-epoxy and glass-epoxycomposites. Mater Des 31:1752–1767
65. Mohan N, Natarajan S, Kumaresh Babu SP, Lee JH (2010) Solidparticle erosion of uhmwpe filled aramid fabric-epoxy hybridcomposites. Adv Mater Res 123–125:1051–1054
66. Fouad A, EI-Meniawi M, Afifi A (2011) Erosion behaviourof epoxy based unidirectional (GFRP) composite materials.Alexandria Eng J 50:29–34
67. Bagci M, Imrek H (2011) Glass fiber reinforced boric acid filledepoxy resin composites. Tribol Int 44(12):1704–1710
68. Bagci M, Imrek H, Khalfan OM (2011) Effects of silicon oxidefiller material and fiber orientation on erosive wear of GF/EPcomposites. World Acad Sci Eng Technol 54:78
69. Patel BC, Acharya SK, Mishra D (2011) Effect of stackingsequence on the erosive wear behavior of jute and jute-glass fab-ric reinforced epoxy composite. Int J Eng Sci Technol 3(1):213–219
70. Kumar S, Satapathy BK, Patnaik A (2011) Thermo-mechanicalcorrelations to erosion performance of short carbon fiberreinforced vinyl ester resin composites. Mater Des 32:2260–2268
71. Powell KL, Yeomans JA, Smith PA (1997) A study of the erosivewear behaviour of continuous fibre reinforced ceramic matrixcomposites. Acta Mater 45(1):321–330
72. Barkoula NM, Gremmels J, Karger-Kocsis J (2001) Dependenceof solid particle erosion on the cross-link density in an epoxyresin modified by hygro thermally decomposed polyurethane.Wear 247:100–108
73. Miyazaki N, Funakura S (1998) Solid particle erosion behaviourof frp degraded by hot water. J Comp Mater 32(13):1295–1305
74. Stachowiak GW, Batchelor AW (1993) Engineering tribology,tribology series 24. Elsevier, Amsterdam, p 586
75. Friedrich K (1986) Erosive wear of polymer surfaces by steel ballblasting. J Mater Sci 21:3317–3332
76. Hutching IM (1992) Ductile-brittle transitions and wear maps forthe erosion and abrasion of brittle materials. J Phys Appl Phys25:212
77. Stack MM, Pungwiwat N (1999) Slurry erosion of metallics,polymers and ceramics: particle size effects. Mater Sci Technol15:337–334
78. Latifi AM (1987). Solid particles erosion in composite materials.Masters Thesis, Wichita State University
79. Gross KJ (1988). Dissertation, Universit¨ at Stuttgart80. Arnold JC, Hutchings IM (1989) Flux rate effects in the erosive
wear of elastomers. J Mater Sci 24:833–83981. Shipway PH, Hutchings IM (1994) A method for optimizing the
particle flux in erosion testing with a gas-blast apparatus. Wear174:169–175
83. Biswas S (2010). Processing, characterization and wear responseof particulate filled epoxy based hybrid composites. Ph.D thesis,NIT Rourkela
84. Barkoula NM, Karger-Kocsis J (2002) Processes and influenc-ing parameters of the solid particle erosion of polymers and theircomposites: a review. J Mater Sci 37(18):3807–3820
86. Rabinowicz E (1979) The wear equation for erosion of metalsby abrasive particles. In: Proceeding of the 5th international con-ference on erosion by liquid and solid impact. Cambridge, pp38–5
87. Finnie I (1958) The mechanism of erosion of ductile metals. In:Proceedings of 3rd US national congress of applied mechanics,pp 527–532
88. Nesic S (1991) Computation of localized erosion-corrosionin disturbed two-phase flow. PhD thesis, University ofSaskatchewan, Saskatoon
89. Bitter JGA (1963) A study of erosion phenomena part I. Wear6:5–21
90. Bitter JGA (1963) A study of erosion phenomena part II. Wear6:169–190
91. Laitone JA (1979) Erosion prediction near a stagnation pointresulting from aerodynamically entrained solid particles. J Air-craft 16(12):809–814
Silicon (2014) 6:5–20 19
92. Salama MM, Venkatesh ES (1983) Evaluation of erosion velocitylimitations of offshore gas wells. In: 15th Annual OTC, May 25,OTC No. 4485. Houston
93. Bourgoyne AT (1989) Experimental study of erosion in divertersystems due to sand production, presented at the SPE/IADCdrilling conference. SPE/IADC 18716, New Orleans, pp 807–816
94. Chase DP, Rybicki EF, Shadley JR (1992) A model for the effectof velocity on erosion of N80 steel tubing due to the normalimpingement of solid particles. Trans ASME J Energy ResourTechnol 114:54–64
95. McLaury BS (1993) A model to predict solid particle ero-sion in oil field geometries. MS Thesis, The University ofTulsa
96. Svedeman SJ, Arnold KE (1993) Criteria for sizing multiphaseflow lines for erosive/corrosive services. Paper Presented at the1993 SPE Conference Houston SPE, 265, 69
97. Jordan K (1998) Erosion in multiphase production of oil andgas, corrosion 98. Paper No. 58. NACE International AnnualConference, San Antonio
98. Shirazi SA, McLaury BS (2000) Erosion modeling of elbows inmultiphase flow. In: Proceedings of 2000 ASME fluids engineer-ing summer meeting, June 11–15. Paper No. FEDSM, Boston,pp 200011251
99. Gomes FC, Ciampini D, Papini M (2004) The effect of inter-particle collisions in international erosive streams on the dis-tribution of energy flux incident to a flat surface. Tribology37:791–807
100. Papini M, Ciampini D, Krajac T, Spelt JK (2003) Computer mod-elling of interference effects in erosion testing: effect of plumeshape. Wear 255(1–6):85–97
101. Gorham DA, Kharaz AH (2000) The measurement of particlerebound characteristics. Powder Technol 112:193–202
102. Wu C, Li L, Thornton C (2003) Rebound behaviour of spheresfor plastic impacts. Int J Impact Eng 28:929–946
103. Karaz AH, Gorham DA (2000) A study of the restitution coeffi-cient in elastic–plastic impact. Philos Mag Lett 80:549–559
104. Molinari JF, Ortiz M (2002) A study of solid-particle erosion ofmetallic targets. Int J Impact Eng 27:347–358
105. Kleis I, Hussainova I (1998) Investigation of particle-wall impactprocess. Wear 233–235:168–173
106. Hutchings IM, Winter RE, Field JE (1976) Solid particle ero-sion of metals: the removal of surface material by sphericalprojectiles. Proc Roy Soc Lond A 348:379–392
107. Hutchings IM (1979) Mechanisms of the erosion of metals bysolid particles. In: Adler WF (ed) Erosion: prevention and usefulapplications, ASTM STP664. American Society for Testing andMaterials, Philadelphia, pp 59–76
108. Hutchings IM (1977) Deformation of metal surfaces by theoblique impact of square plates. Int J Mech Sci 19:45–52
109. Rickerby DG, MacMillan NH (1980) On the oblique impact of arigid sphere against a rigid plastic solid. Int J Mech Sci 22:491–494
110. Hutchings IM, Macmillan NH, Rickerby DG (1981) Furtherstudies of the oblique impact of a hard sphere against a ductilesolid. Int J Mech Sci 23(11):639–646
111. Sundararajan G, Shewmon PG (1987) The oblique impact ofa hard ball against ductile, semi-infinite, target materials—experiment and analysis. Int J Impact Eng 6(1):3–22
112. Tirupataiah Y, Venkataraman B, Sundararajan G (1990) Thenature of the elastic rebound of a hard ball impacting on ductile,metallic target materials. Mater Sci Eng A 24:133–140
113. Sundararajan G (1991) The depth of the plastic deformationbeneath eroded surfaces: the influence of impact angle and
velocity, particle shape and material properties. Wear 149:129–153
114. Papini M, Spelt JK (1998) The plowing erosion of organiccoatings by spherical particles. Wear 222:38–48
115. Papini M (1999) Organic coating removal by single particleimpact. PhD thesis, Department of Mechanical and IndustrialEngineering, University of Toronto, Toronto, Ont., Canada
116. Papini M, Spelt JK (2000) Impact of rigid angular particles withfully plastic targets—part, I., analysis. Int J Mech Sci 42(5):991–1006
117. Papini M, Spelt JK (2000) Impact of rigid angular particleswith fully plastic targets—Part II: parametric study of erosionphenomena. Int J Mech Sci 42(5):1007–1025
119. Hutchings IM (1981) A model for the erosion of metals byspherical particles at normal incidence. Wear 70:269–281
120. Brown R, Jun E, Edington J (1982) Mechanisms of solid particleerosive wear for 908 impact on copper and iron. Wear 74:143–156
121. Patnaik A, Satapathy A, Mahapatra SS, Dash RR (2009) Tribo-performance of polyester hybrid composites: damage assessmentand parameter optimization using taguchi design. Mater Des30:57–67
122. Patnaik A, Satapathy A, Mahapatra SS, Dash RR (2008) Mod-eling and prediction of erosion response of glass reinforcedpolyester-flyash composites. J Reinf Plast Compos 28:513–536
123. Patnaik A, Satapathy A, Mahapatra SS, Dash RR (2008) Acomparative study on different ceramic fillers affecting mechan-ical properties of glass-polyester composites. J Reinforc PlastCompos 28:1305–1318
124. Patnaik A, Satapathy A, Mahapatra SS, Dash RR (2008) Erosivewear assesment of glass reinforced polyester-flyash compositesusing taguchi method. Int Polym Process 13:192–199
125. Mahapatra SS, Patnaik A, Satapathy A, Dash RR (2008)Taguchi method applied to parametric appraisal of erosionbehaviour of gf-reinforced polyester composites. Wear 265:214–222
126. Biswas S, Satapathy A (2009) Tribo-performance analysis of redmud filled glass-epoxy composites using taguchi experimentaldesign. Mater Des 30:2841–2853
127. Phadke MS, Dehnad K (1988) Optimization of product and pro-cess design for quality and cost. Qual Reliab Eng Int 4(2):105–112
128. Hovis SK, Talia JE, Scattergood RO (1986) Erosion in multi-phase systems. Wear 108:139–155
129. Ballout YA, Hovis SK, Talia JE (1990) Erosion in glass–fiberreinforced epoxy composite. Scr Metall Mater 24:195–200
131. Miyazaki N, Takeda N (1993) Solid particle erosion of fiberreinforced plastics. J Compos Mater 27:21–31
132. Hager A, Friedrich K, Dzenis YA, Paipetis SA (1995) Study onerosion wear of advanced polymer composites. In: Street K (ed)Proceedings of ICCM-10, Whistler, B.C., Canada. WoodheadPublishing Ltd., Cambridge, pp 155–162
133. Marei AI, Izvozchikov PV (1967) Determination of the wear ofrubbers in a stream of abrasive. Abrasion of rubber. MacLaren,pp 274–280
134. Besztercey G, Karger-Kocsis J, Szaplonczay P (1999) Solid par-ticle erosion of electrically insulating silicone and epdm rubbercompounds. Polym Bull 42:717–724
20 Silicon (2014) 6:5–20
135. Fernandez JE, Fernandez MDR, Diaz RV, Navarro RT (2003)Abrasive wear analysis using factorial experiment design. Wear255(1–6):38–43
136. Spuzic S, Zec M, Abhary K, Ghomashchi R, Reid I (1997) Frac-tional design of experiments applied to a wear simulation. Wear212(1):131–139
137. Prasad BK (2002) Abrasive wear characteristics of a zinc-basedalloy and zinc-alloy/sic composite. Wear 252(3–4):250–263
138. Deuis RL, Subramanian C, Ellup JM (1998) Three-body abrasivewear of composite coatings in dry and wet environments. Wear214(1):112–130
139. Banerji A, Prasad SV, Surappa MK, Rohatgi PK (1982) Abrasivewear of cast aluminium alloy-zircon particle composites. Wear82(2):141–151
140. Mondal DP, Das S, Jha AK, Yegneswaran AH (1998) Abrasivewear of al alloy–al2o3 particle composite: a study on the com-bined effect of load and size of abrasive. Wear 223(1–2):131–138
141. Taguchi G, Konishi S (1987) Orthogonal arrays and lineargraphs: tools for quality engineering. American Supplier InstituteInc., Dearborn, p 72
142. Taguchi G (1986) Introduction to quality engineering: designingquality into products and processes. Asian Productivity Organi-zation, Tokyo
143. Phadke MS (1989) Quality engineering using robust design.Prentice-Hall, Englewood Cliffs
144. Wu Y, Moore WH (1986) Quality engineering: product &process design optimization. American Supplier Institute Inc.,Dearborn
145. Shoemaker AC, Kacker RN (1988) A methodology for planningexperiments in robust product and process design. Qual ReliabEng Int 4(2):95–103
146. Mahapatra SS, Patnaik A, Khan MS (2006) Development andanalysis of wear resistance model for composites of aluminiumreinforced with red mud. J Solid Waste Tech Manag 32(1):28–35
147. Mahapatra SS, Patnaik A (2006) Optimization of parameter com-binations in wire electrical discharge machining using taguchimethod. Indian J Eng Mater Sci 13:493–502
148. Mahapatra SS, Patnaik A (2006) Optimization of wire electricaldischarge machining (WEDM) process parameters using taguchimethod. Int J Adv Manuf Technol 34(9–10):911–925
149. Mahapatra SS, Patnaik A (2007) Parametric optimization of wireelectrical discharge machining (WEDM) process using taguchimethod. J Braz Soc Mech Sci 28(4):423–430
150. Mahapatra SS, Patnaik A (2006) Determination of optimal para-meters settings in wire electrical discharge machining (WEDM)process using taguchi method. J Inst Eng (India) 87:16–24
151. Mahapatra SS, Patnaik A (2006) Parametric analysis and opti-mization of drilling of metal matrix composites based on thetaguchi method. Int J Manuf Sci Prod 8(1):5–12
152. Mahapatra SS, Patnaik A (2007) Optimization of wire electricaldischarge machining (wedm) process parameters using taguchimethod. J Manuf Sci Tech 9(2):129–144
153. Patnaik A, Satapathy A, Mahapatra SS, Dash RR (2010) Mod-ified erosion wear characteristics of glass-polyester compositesby silicon carbide filling: a parametric study using taguchitechnique. Int J Mater Prod Technol 38(2–3):131–152
154. Sarle WS (1997) Neural network FAQ: part 1 of 7—introduction,periodic posting to the use net news group Comp.ai.neural-nets.Available from: ftp://ftp.sas.com/pub/neural/FAQ.html
155. Velten K, Reinicke R, Friedrich K (2000) Wear volume predic-tion with artificial neural networks. Tribol Int 33(10):731–736
156. Zhang Z, Friedrich K, Velten K (2002) Prediction on tribologi-cal properties of short fiber composites is using artificial neuralnetworks. Wear 252(7–8):668–675
157. Kang JY, Song JH (1998) Neural network applications in deter-mining the fatigue crack opening load. Int J Fatig 20:57–69
158. Malinova T, Malinov S, Pantev N (2001) Grinding mode iden-tification and surface quality prediction using neural networksin grinding of silicon nitride. Surf Coat Technol 135:258–267
159. Zeng Q, Zu J, Zhang L, Dai G (2002) Designing expert systemwith artificial neural networks for in situ toughened Si3N4. MaterDes 23:287–290
160. Yescas MA, Bhadieshia HKDH, MacKay DJ (2001) Estimationof the amount of retained austenite in austempered ductile ironsusing neural networks. Mater Sci Eng Part A 311:162–173
161. Jain RK, Jain VK, Kalra PK (1999) Modelling of abrasive flowmachining process: a neural network approach. Wear 231:242–248
162. Zhang Z, Friedrich K (2003) Artificial neural network appliedto polymer composites: a review. Compos Sci Technol63(14):2029–2044
