Professional Documents
Culture Documents
PII: S0263-8223(16)30767-X
DOI: http://dx.doi.org/10.1016/j.compstruct.2016.10.015
Reference: COST 7841
Please cite this article as: Giasin, K., Ayvar-Soberanis, Sabino., An Investigation of Burrs, Chip formation, Hole
Size, Circularity and Delamination during Drilling Operation of GLARE using ANOVA, Composite Structures
(2016), doi: http://dx.doi.org/10.1016/j.compstruct.2016.10.015
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Article title: An Investigation of Burrs, Chip formation, Hole Size, Circularity and Delamination during
Drilling Operation of GLARE using ANOVA.
While there has been a significant amount of research on drilling composite-metal stacks, limited work has been
carried out on the machinability of fibre metal laminates (FMLs) used in aerospace structures. Challenges faced
in drilling FMLs include those which exist in metals and composites such as delamination, chip and burr
formations. The aim of this work is to extend the knowledge of machining fibre metal laminates through the
assessment of twist drilling operations in order to improve workpiece quality. The current work presents an
experimental study to analyse the effects of drilling parameters (spindle speed and feed rate) on hole quality in
two grades of GLARE (2B & 3). The evaluation includes inspecting the hole size, circularity error, entry and
exit burrs, chip formations and damage described at the macro level (delamination area) using computerised
tomography CT scan, and at the micro level (fibre matrix debonding, chipping, adhesions, cracks) using
scanning electron microscopy (SEM). In addition, the results are statistically analysed using analysis of variance
(ANOVA) to determine the contribution of cutting parameters on investigated hole quality parameters.
1.0 Introduction
Glass ALuminuim Reinforced Epoxy (GLARE) is a hybrid metal-composite material which consists of
alternating layers of Al2024 sheets and glass fibre layers bonded together using FM94 film adhesive. It is
currently installed in the front and rear parts of the fuselage in the Airbus A380 commercial aircraft [1]. GLARE
is considered to be one of the most promising aerospace materials as it offers 10% weight reductions with
respect to monolithic aluminium [2]. 25% of the Airbus A380 airframe is made of composites, 22% of which
are carbon or glass fibre reinforced plastics CFRPs and 3% of which are GLARE [3]. The assembly of the
aircraft structural parts is carried out using fasteners and riveted bolts, which requires drilling a large number of
holes. The number of holes required may vary from 300,000 holes in a jet fighter up to 1.5-3 million holes in
commercial aircraft [4]. For example, a single wing in Airbus A380 requires the drilling of 750,000 holes [5].
Extra care is taken to avoid misalignments during the drilling process as poor hole quality can result in cracks
within the airframe structure, which can reduce their service lifetime and add extra costs for maintenance.
Therefore, drilling requires the use of proper cutting parameters to improve the surface finish of the hole, while
drilling GLARE laminates requires different cutting parameters from those used for its constituents, aluminium
focused on evaluating the cutting forces and hole quality parameters such as surface roughness, burr formation
and drilling/milling induced damage such as delamination [2, 6, 8, 9]. Early literature during the development of
GLARE reported that the feed rate, lubrication and laminate thickness in milling operations had a major
influence on the quality of the milled edge [2]. In milling, the tool wear and fracture were found to increase with
the increase in feed rate which also reduced the laminate edge quality and caused delamination [2]. Previous
investigation on the cooling/lubricant influence on milling GLARE showed that using coolant is preferable over
semis dry cooling especially when machining thick laminates due to the limited ability of the semi-dry cooling
to effectively remove the heat generated during the milling process [2], while using cryogenic and minimum
quantity lubrication coolants during drilling of GLARE showed the ability to reduce exit burr and hole
circularity error compared to dry drilling at room temperature [11]. The use of coated carbide cutting tools over
non-coated and high-speed steel (HSS) tools is recommended for better wear resistance and hole quality due to
the abrasive nature of the glass fibres in GLARE [6]. The studies also looked into the effect of tool size
geometry on developed cutting forces and hole quality [6, 9] and found that two facet drill outperformed the
other cutting tools in terms of eliminating delamination and acceptable burr formation, while drilling GLARE
using four and eight facet drills produced poor hole quality and uncut fibres around the edges of machined holes
[9]. Table 1 shows a summary of the most recent research on drilling fibre metal laminates.
Table 1: Summary of the most recent research on drilling GLARE fibre metal laminates
Twist drills are widely used in industry to produce holes rapidly and economically [15]. They are often
produced smaller than the stated nominal size and they are commonly used to produce riveted holes [16]. For
aerospace structures, riveting and bolting typically require tolerances as low as ±0.025 mm [17]. Achieving
holes with tight tolerances is governed by several factors such as the mechanical properties of the materials
itself, its hardness, its coefficient of thermal expansion and conductivity. Additionally, cutting tool rigidity and
the actual size of the tool. Delamination is a specific failure mode for GLARE laminates and is most likely to
occur on top or the bottom layers when drilling [2]. Burr formations are common when drilling GLARE
laminates, they represent unnecessary material that remains after machining and its removal requires deburring
which is usually done manually because of difficulties in automation [18]. Although burr height is more
commonly measured characteristic, burr thickness contributes more to deburring costs than burr height [19].
Deburring accounts for about 30% of the total manufacturing cost [20] and it occupies more than 40% of the
total machining time and thereby reduces production efficiency [21]. Deburring in GLARE is cumbersome
because burrs can form internally within the metal layers of the hybrid stack known as interlayer burrs as well as
at the entrance and exit sides. The raised burrs formed from the metal sheets may penetrate the composite layers,
thereby reducing the overall quality of the machined part. Table 2 depicts some of the reported burr height and
circularity error ranges in previous drilling trials on GLARE, CFRP/metal stack, CFRPs and aluminium alloys.
Table 2: Some of the reported hole circularity error, burr height and burr root thickness during drilling GLARE,
composite metal stacks, composites and aluminium alloys
Analysing chip formation during the machining process can relate to the cutting mechanism the material
undergoes when it is deformed and removed ahead of the cutting tool [32]. The analysis is important since
formed chips influence the surface finish of machined parts. The effect of chip formation becomes critical when
drilling composite-metal stacks and FMLs since the evacuated chips continuously rub against the internal wall
of the hole eroding the prepreg plies throughout the thickness and possibly cause delamination. Therefore, the
impact of spindle speed and feed rate during drilling two grades of GLARE are experimentally investigated and
statistically analysed by the analysis of variance (ANOVA). The aim is to evaluate the impact of these cutting
parameters on the hole quality in GLARE laminates. The investigation of hole quality includes measuring the
hole size, circularity error, burr formations at the upper and lower surface of the outer aluminium sheets and
chip formation. computerised tomography scan (CT-scan) and scanning electron microscopy (SEM) techniques
are utilised to inspect the glass fibre layers for delamination -if any- and defects in the borehole surface.
Figure 1: Schematic illustrations of GLARE laminates: (a) GLARE 2B with the unidirectional fibre coinciding
with the rolling direction [14] (b) GLARE 3 with the cross-plied fibre layers [14] (c) Front view of GLARE 2B
11/10-0.4 workpiece [10] (d) Front view of GLARE 3 8/7-0.4 workpiece [10] (e) Top view of the workpiece
[10]
The drilling trials were carried out using MORI SEIKI SV-500 CNC milling machine with a maximum spindle
speed of 10000 rpm. Two grades of GLARE were used in this investigation: GLARE 2B & GLARE 3 as shown
in Figure 1. The samples were provided by the Fibre Metal Laminate Centre (FMLC) in Netherlands. Additional
details on the samples are provided in Table 3. Each GLARE sample consisted of sheets of Al2024-T3
aluminium alloy having a nominal thickness of 0.4064 mm and prepregs of S2-glass fibres embedded in an
FM94 adhesive having an approximate thickness of 0.133 mm [2, 8, 10, 12-14]. The samples were cured in an
autoclave for around 300 minutes at elevated temperatures of 120 °C and under a pressure of 6 bars [33]. The
fibres are delivered as a prepreg including the FM94 adhesive system from Cytec in the U.K [2, 8, 10, 12-14].
The cutting tools used for all drilling tests were Φ 6mm OSG HYP-HP-3D TiAlN coated carbide twist drills
Table 3: GLARE grades used in the drilling experiments and their properties [8, 10, 12]
The cutting parameters used in the current study are summarised in Table 4. Four spindle speeds and four feed
rates were used, which represent a full factorial design with two factors and four levels (42= 16 holes). In order
to confirm the repeatability of the experiment, the tests were repeated two additional times and results reported
here are the average from three tests. Each set of 16 holes was drilled with a new tool to avoid any effects that
might arise from the tool wear and no coolants were used in this study [8]. As shown previously in the literature,
there were limited studies available on drilling GLARE. Generally, the feed rates used in drilling GLARE,
composite metal stacks, aluminium alloys and glass fibre reinforced plastics (GFRP) was between 0.05 to 0.3
mm/rev while the spindle speeds -depending on the size of the cutting tool- ranged between 4000 to 7500 rpm
Table 4. Spindle speeds and feed rates used in the experimental work.
hole entry and exit sides. The parameters were measured using a surface roughness profilometer as shown in
Figure 2, using burr parameters described by [35]. The full details of the measurement process were described in
Figure 3: Measurement setup of hole size and circularity error using CMM machine and description of MIC
method [10, 37]
Circularity, also known as roundness is a two dimensional radial tolerance that describes how close is a part
with a diametrical cross section to a true circle. It is used as a tool to control the amount of maximum allowable
circularity error (roundness error) of the circular part internally such as in holes or externally such as shafts and
bearings [38]. Hole size and circularity error were measured using SHEFFIELD CORDAX D8 CMM machine
shown in Figure 3. The full details of the measurement process were summarised in details in one of our
previous studies [10]. The workpiece is placed on the machine table, and a ruby probe (2 mm diameter)
continuously circulate around the internal wall of the hole. The measurements were carried out at two locations
along the hole depth stated as ‘top’ and ‘bottom’ as shown in Figure 4. The measurements were taken at 1 and 6
mm below the hole entry side for Glare 2B 11/10 and at 1 and 4 mm for GLARE 2B 8/7 and Glare 3 8/7. The
scanning speed for the measuring probe was set at 1 mm per second which allows capturing 400 points while
rotating around the inner circumference of a hole. The MIC method was used for measuring the circularity error
in the current experiments (see Figure 3) which can be calculated using the following equation:
Where
• Rmax : which is the maximum radius of a circle that can be circumscribed outside the circular hole
profile.
• Rmin : which is the maximum radius of a circle that can be inscribed and completely enclose the circular
hole profile without cutting it and is concentric with the maximum circumscribed circle.
Figure 4: the top and bottom locations of the measurements of hole size and circularity error in GLARE
laminates used in the current study
2.2.3 Delamination
The glass fibre layers were inspected for drilling induced delamination using X-ray computerised tomography
(CT) scanning technique using Nikon Metrology’s XTH 225 ST and Nikon Metrology XTH 225/320 LC CT
scan machines [13]. The samples are stacked onto a turntable using floral foam material as shown in Figure 5.
The florist foam is commonly used in CT scans due to its low density and reasonable rigidity which makes it
relatively transparent to X-rays. The X-ray voltage and current were set at 135 kV and 105 mA respectively
[13]. The sample exposure time was set to 1000 ms per projection with a total number of 3300 projections and a
scanning resolution of 15 microns. Nikon Metrology Inspect-X and XT CT Pro/CT agent were used for setting
up the scanner and reconstruction of CT volume [13]. Volume graphics Gmbh (VG 2.2) software was used for
analysis of the reconstructed volume and to capture the images for delamination analysis of the composite
layers.
The SEM technique was employed to analyse drilling induced damage to the bore wall surface. Each hole was
cross-sectioned from its centre and was cleaned using acetone in an ultrasonic bath for ten minutes to remove
glass fibre dust and any debris on the surface of the borehole as shown in Figure 6.a. [13]. The samples were
then placed on the top of a carbon sticker and inserted inside the SEM chamber for surface inspection as shown
in Figure 6.b. The SEM scanning was conducted using Hitachi HM3030 plus tabletop microscope as shown in
Figure 6.c.
Figure 6. (a) SEM sample preparation (b) GLARE samples mounted inside the SEM chamber (c) Hitachi
TM3030 table top SEM microscope
Figure 7 show a comparison of the average hole size at top and bottom locations under different spindle speeds
and feed rates. Overall, the variation of the hole size ranged between -38.3 and +18.7 µm of the drill nominal
size. Generally, holes drilled in GLARE 3 8/7 and GLARE 2B 8/7 were oversized at the top, while the drilled
holes in GLARE 3 8/7 and GLARE 2B 11/10 were undersized at the bottom. The reduction in hole size in
GLARE 3 8/7 at the bottom was greater than in GLARE 2B 8/7 and 11/10, which indicate that the hole size
could be affected by the fibre orientation in the laminate. Additionally, the shrinkage of the hole size at the
bottom increased with the increase of spindle speed due to the rise in workpiece temperature during the drilling
process.
The glass fibre-epoxy matrix and metal sheets in GLARE react differently to the rise of workpiece temperature
in the cutting zone. In metal machining, up to 70% of the heat generated at the cutting zone is carried away by
chips [32], while the relatively low thermal conductivity of glass fibre layers means that the generated heat does
not conduct quickly to the cutting tool. The heat transfer occurs only in a thin layer at the interface between the
aluminium sheets and the prepregs throughout the laminate. Additionally, the mismatch coefficient of thermal
expansion of GLARE constituents can lead to residual stresses and dimensional variations in the hole diameter
[32]. The thermal expansion coefficient in unidirectional plies vary with direction, for example, the thermal
expansion coefficient of a cross-ply in fibre direction α1 = 6.1· 10-6 1/ºC while the thermal expansion in the
transverse or out of plane direction α2 = 26.2·10-6 1/ºC [39] which is due to the transverse isotropy of the glass
fibre epoxy [40]. The heat generated during the drilling process followed by cooling of the workpiece causes
thermal expansions and shrinkages of the glass fibre layers. The thermal expansion coefficient of glass fibre
layers is different relative to the fibre direction, the contraction in glass fibre layers in cross-ply configuration
acts in different directions. Therefore, the retraction to its original position will be slower than that in
unidirectional ply configurations. The 0° fibres expand in a longitudinal direction while the 90° fibres contract
in a transverse direction on the same axis which leads to geometric mismatch and residual stresses in glass fibre
layers [41].
The geometric mismatch influences the dimensional accuracy of the drilled hole at the glass fibre and causes
them to have a smaller diameter than the nominal diameter of the drill bit. Undersized holes are common when
drilling laminated composites [42] due to the elastic bending and tightening of the reinforcement fibres during
machining followed by relaxation of the lamina [43]. This was evident by the fact that the reduction in hole size
at a bottom location in GLARE 2B 11/10 was greater than that in GLARE 2B 8/7, which indicates that hole size
is likely to shrink with depth. In GLARE 3 8/7 and GLARE 2B 8/7, the hole size at the top increased with the
increase of the spindle speed and decreased with the increase of the spindle speed at the bottom. The results
were analysed using ANOVA statistical tool and MINITAB 17 and the percentage contribution of each cutting
parameter and their interaction are provided in Table 5. Results show that the influence of spindle speed was
more significant on hole size than the feed rate when drilling GLARE 2B 11/10 and GLARE 2B 8/7 at the top.
The interaction between the spindle speed and the feed rate was insignificant for both GLARE 2B thicknesses at
the bottom. The error in the model was large which could be due to the presence of quadratic terms in that
affects the input parameters. The influence of the feed rate and the spindle speed on hole size at the bottom were
approximately equal in GLARE 3 8/7 with 22.11% for the feed rate and 21.31% for spindle speed while their
interaction significantly contributed by 30.62%. Hole size at the top showed that both cutting parameters and
their linear interaction were insignificant which could be due to the presence of quadratic terms in the model
that affect the input parameters. Therefore, the model requires further statistical analysis which will be carried
out in a future study. The large error percentage in hole size and circularity error is due to the large variation
among the individual observations in data of the repeated measures which indicates a poor estimate of the mean.
Table 5: Percentage contribution of cutting parameters on the hole size, circularity error, burr height and burr
thickness.
Figure 8 show the circularity error measured at different spindle speeds and feed rates for the tested GLARE
grades at top and bottom locations. Overall, it is observed that the circularity error increased with the increase of
spindle speed due to the increase of chatter vibrations that causes instability of the cutting tool. At higher spindle
speeds the cutting tool rubs around the hole wall more frequently causing higher distortions which increase the
circularity error. Additionally, it is observed that the circularity error increased with the increase of the feed rate
which could be due to the increase in cutting forces. The faster penetration of the cutting tool through the
workpiece -due to the increase of the feed rate- increases the hole deflections and vibrations in the cutting tool
causing higher circularity errors. The circularity error increased with hole depth (workpiece thickness) which
could be due to the increase in frictional forces, heat and vibrations with depth [44]. The increase in circularity
error at the bottom compared to the top was greater in GLARE 3 8/7 than in GLARE 2B 8/7 which indicates
that the fibre orientation has an influence on circularity error. The circularity error was higher in GLARE 2B for
spindle speeds of = 1000 and 3000 rpm and lower for spindle speeds of = 6000 and 9000 rpm in
1.03 to 104 µm at the bottom. The circularity error at top and bottom in GLARE 2B 11/10 were greater than
those in GLARE 2B 8/7 which could indicate that the circularity error increase with laminate thickness.
ANOVA analysis (see Table 5) indicate that the feed rate and the spindle speed had slightly higher impact on
the hole circularity error when drilling GLARE 2B 11/10 with 18.64 % and 13.34 % for spindle speed and feed
rate respectively, while their interaction had a significant contribution of 27.02% and similar to hole size the
error in the model was large. For GLARE 2B 8/7, the influence of the spindle speed was more significant than
the feed rate with 24.45 % and 27.92 % at top and bottom respectively. The feed rate contribution was around
14.5 %. The interaction between the spindle speed and the feed rate seemed to have a significant influence
which ranged between 34.08 to 41.43 % which indicates that the influence of spindle speed becomes more
significant with depth. For GLARE 3 8/7, the influence of the spindle speed on hole size at the bottom was
approximately higher with 20.73 % for the feed rate and 13.88 % for feed rate while their interaction
significantly contributed by 36.04 %. Hole circularity error at the top showed that both cutting parameters and
their linear interaction were insignificant which could be due to the presence of quadratic terms in the model
that affect the input parameter, this is evident by the large error present in the model 66.19 %. Therefore, the
model requires further statistical analysis which will be carried out in a future study.
To conclude on hole size and circularity error analysis and results during dry drilling of GLARE laminates,
typical hole demands when drilling composite metal stack materials commonly requires to achieve an H7 hole
tolerance fit based on the ISO 286 standard. This means that the hole should not vary more than ±12 microns but
more relaxed tolerances such as H8 (±18 microns) or H9 (±30 microns) were allowed to be used when drilling
composites due to difficulty in achieving H7 tolerances, for example, SANDVIK recommends that drilled hole
size tolerance in a composite metal stack should be between ±20 and ±40 microns [13, 45]. The hole circularity
error ranged between 6 and 35 microns in all GLARE grades and thicknesses under most cutting parameters.
Generally, smaller circularity error is preferred and since there was no specific acceptable standard for hole
circularity error, comparing the results with circularity error results from previous literature shown earlier (see
Table 2), circularity error could range anything from 6 µm and as high as 312 µm, the results in the current
study were similar to those reported by Zitoune et al. on machining composite metal stack materials [25].
3.3 Effect of cutting parameters on burr height and burr root thickness
Figure 9 shows SEM and optical microscopy images of the edge of the first and last aluminium sheets in a hole
drilled in GLARE 11/10-0.4. The exit burr was considerably larger than entrance burr. The edges of the last
aluminium sheet in all drilled holes were more deteriorated and deformed than the first one due to the increased
cutting temperatures with depth which rise the ductility of the metal causing greater deformations during the
drilling process. Entry burr size is small and the drilled holes can be considered burr free at the entrance. All the
burrs formed -with the exception of the holes drilled at high feed rates and low spindle speed of =600 and 900
mm/min and =1000 rpm- were very small and mostly uniform burr type. Some of the formed burrs could not
be seen by the naked eye as shown in Figure 9 c and d. This type of burr does not exhibit evidence of material
that can break away and is generally acceptable in machining operations and does not require deburring unless it
can affect the assembly of holes in mating parts or the surface of surface contact.
Figure 9: SEM images of (a) entrance burr and (b) exit burr in GLARE 2B 11/10-0.4 at = 9000
and
= 100
/
(c) exit burr formation for hole drilled at 9000 rpm and 900 mm/min (d) exit burr formation
for hole drilled at 3000 rpm and 300 mm/min
Figure 10 and Figure 11 show the entrance and exit burr heights and root thicknesses under different spindle
speeds and feed rates. For all GLARE grades and thicknesses tested, the average burr height increased with the
increase of the feed rate on both sides. The influence of spindle speed on burr height varied depending on its
level and the applied feed rate. Several burr caps were formed in each hole but they tended to separate when the
tool exited the workpiece. Burrs were small and could be only observed using a microscope. Additionally,
results showed that the burr heights in GLARE 2B 8/7 were greater than those formed in GLARE 3 8/7 on both
sides which could indicate that fibre orientation can influence the burr formation mechanism. The results of burr
height and thickness for = 600 and 900 mm/min feed rate and =1000 rpm spindle speed are not shown in
the figures due to their large value >300 µm. The influence of increase the feed rate on entry burr height was
greater in GLARE 2B than that in GLARE 3. This could be due to the fact that the energy transfer from one
composite layer to another is easier if they both have same fiber orientation such as in GLARE 2B and as a
results the severity of the damage is greater than what would be if the fibres in each layer had different
orientations such as in GLARE 3 and allow it to fail at higher loads [46]. Conversely, the entry burr thickness
did not show any significant relationship with fibre orientation, nevertheless, the exit burr thickness was higher
and thickness. The contribution of spindle speed on burr formation ranged between 28.57 to 43.45 %, followed
by the feed rate with 16.89 to 23.87 % contribution. The interaction between the spindle speed and the feed rate
had a significant contribution which ranged between 38.76 to 49.75 %. The burr height formed in GLARE 2B
8/7 was generally higher on both sides than that in GLARE 2B 11/10. This was also observed when comparing
GLARE 3 8/7 against GLARE 2B 11/10, the burr height was slightly higher at entry. Burr height decreased with
workpiece thickness and the maximum burr height at entry was recorded at = 3000 rpm and = 900 mm/min
(57 m/min and 0.3 mm/rev). The burr root thickness in GLARE 2B was higher than in GLARE 3 8/7 at the exit
side and it increased with workpeice thickness at both sides. This indicates that the fibre orientation and
workpiece thickness played an important role in determining the size and height of burrs. The higher burr root
thickness at exit in GLARE 2B than GLARE 3 could be also attributed to higher temperatures developed at the
exit side of the workpiece during the drilling process. It was previously reported in one of our previous studies
that the developed temperatures at the bottom surface of GLARE 2B are likely to be higher than in GLARE 3
Figure 11: Average burr root thickness at (a) entrance (b) exit
Interlayer burrs were formed in the drilled holes as shown in Figure 17 but they were difficult to measure. The
interlayer burrs were not uniform around the edge and formed randomly around the hole periphery by
protruding glass fibre layers. The interlayer burrs are usually formed when the elastic bending deformation of
two different materials stacked together occurs at different degrees due to the mismatch in physical and
mechanical properties. In addition, while the cutting tool drills through the stack, the upper part of the
workpiece undergoes elastic recovery while the lower part undergoes large elastic bending deformation due to
the acting of the feed force. The bending of the workpiece allows for the formation of small gaps which
provides suitable conditions for interlayer burr formation. The formed chips can be also forced onto the glass
fibre layers surfaces while the cutting tool is cutting through or retracting from the workpiece. Therefore, to
conclude on burr height and burr root thickness analysis during dry drilling of GLARE laminates, as reported
from the literature, there is no available data on the acceptable burr height or thickness for GLARE, composite
metal stacks or monolithic aluminium for machining/drilling process. However, comparing the obtained data
with previous studies on drilling aluminium alloys and GLARE fibre metal laminates shows that exit burr height
could reach anything between 10 to 80 µm [9, 28-31] . Generally, the burr height obtained from all drilling trials
are within or below those obtained in previous studies on drilling GLARE and aluminium alloys.
/
In order to investigate the detailed morphology of the machined holes, scanning electron microscopy was used
to examine the borehole surface of the samples as shown in Figure 12.a. Prior cleaning the drilled holes, some
samples were inspected under the microscope as shown in Figure 12.b. Separation of tiny bundles of fibres from
the composite layers was observed on the walls of the drilled holes which indicates that powdery chips are likely
to form when drilling GLARE. It was observed that the damage and deterioration in hole surface increased with
the increase of the feed rate and the spindle speed. The feed rate seemed to cause a larger impact on the inside
walls of the holes which could be due to the action of the thrust force which pushes down the glass fibre layers
causing bending and deformations. The aluminium sheets on the top and bottom of each glass fibre layers act as
a backup plate which helped reduce delamination and damage in the layers. Tsao et al. [47, 48] previously
reported that the existence of backup plate at the exit side of drilled hole in composites allows for larger critical
thrust force (i.e larger feed rates) to be used without delamination damage. The reduction is attributed to the
suppression of crack growth by the upward reaction of the backup plate [49]. In the case of GLARE, the glass
fibre layers are supported on both sides by metal sheets which provide a form of resistance to peeling up and
push out delamination common in drilling composites. Best surface hole quality was achieved when drilling at
feed rates of = 100 mm/min and spindle speeds of = 1000 and 3000 rpm.
Figure 13: Good hole quality in (a) GLARE 2B 8/7 under = 6000
and = 600
/
(0.1 mm/rev)
(b) GLARE 2B 11/10 under = 3000
and = 100
/
(0.03 mm/rev)
It was also observed that the damage in GLARE 3 8/7 was more severe than in GLARE 2B 8/7. Interlaminar
failure was observed in the middle glass fibre layer of some of the hole drilled in GLARE 3 8/7. A critical factor
that affects the damage tolerance in composites is the fibre direction. Generally, the energy can be easily passed
from one composite layer to the other if they both have same fibre orientation and as a result, the severity of
damage is greater than what would be if the fibres in each layer had different orientations, since this will have a
transfer of energy between layers and allow it to fail at a higher load. Therefore, from a crack resistance point of
view, a [0/90/0/90] stacking sequence is preferred over [0/0/90/90] or [0/90/90/0] [46]. However, from a
separation resistance point of view, the interlaminar interface between laminates with different ply orientations
(i.e. cross-ply configuration) are mechanically weak and therefore, local separation of the laminate from one
another is a common form of damage in such configurations [50]. This would imply that the influence of feed
rate on GLARE 3 is more critical than that on GLARE 2B laminates. The interlaminar failure in GLARE 3
could be also related to the mismatch in coefficient of thermal expansion (CTE) between plies of different
orientations which induces microcracks and thermally induced strains that alter the thermal and mechanical
/
Under some specific cutting parameters, the borehole surface can be damage-free (see Figure 13), while under
some other cutting parameters, increasing the feed rate or spindle speed lead to poor surface quality causing
fibre damage, smearing, built-up edge, and in some cases, separation between glass fibre layers and debonding
between the glass fibre layers and aluminium sheets. The SEM analysis revealed that the subsurface damage
increased with hole depth as shown in Figure 14. The last few aluminium sheets and glass fibre layers in
GLARE 2B 11/10 were more damaged (see Figure 14.a) and adhesions were more visible than in that in
GLARE 2B 8/7 (see Figure 14.b). The increase in spindle speed showed somewhat better fibre cutting and less
protruding and uncut fibres. However, drilling under spindle speeds of = 9000 rpm showed debris (waste)
adhered to the walls of the hole as shown in Figure 14 regardless of the feed rate used and was more apparent in
thicker GLARE 2B 11/10 samples than in GLARE 2B 8/7 or GLARE 3 8/7. Drilling at high spindle speeds
increase the cutting temperatures, while drilling thicker laminates gives larger contact area between the cutting
tool and the workpiece, which produces more heat. At higher temperatures the chip and the fibres adhere to the
cutting tool which can be forced into the laminate edges and borehole surfaces leading to undesirable
deformations and delamination and creating what is called waste material [2]. The waste material can lead to
/
(c) (d) GLARE 2B 11/10 at =
3000
, = 300
/
The CT scan inspection of surface delamination in glass fibre layers revealed that it is possible to obtain a
surface delamination free holes when drilling at proper spindle speed and feed rate combinations. The results of
the analysis show little or no surface delamination for both grades of GLARE on all glass fibre layers due to the
layered structure of GLARE which stacks glass fibre layers between two metallic sheets. As mentioned earlier,
Tsao and Abrao previously reported that the use of a backup plate can significantly reduce the delamination by
providing support and minimising deformations [47, 48]. In addition, no microcracks were found through the
thickness or the transverse plane of the layers in drilled holes except when drilling at feed rates of = 600 and
900 mm/min and spindle speed of = 1000 rpm (see Figure 15.a and 15.b).
Figure 16: An example of (a) fibre buckling in drilling GLARE 3 8/7 (b) Fibre length variation with angle of cut
in GLARE 2B 8/7 (c) SEM image of the machined surface resulting from drilling GLARE 3 8/7 showing fibres
partly embedded in the epoxy matrix and cracks across the fibres at 9000 rpm and 300 mm/min
In GLARE 3 (0/90 ply orientation), it was observed that the delamination type chip formation (Type I) which
usually occurs in the 0° fibre orientation was present on the inspected glass fibre plies. The mode I fracture
occurs due to the advancement of the cutting tool into the workpiece which causes the peeled glass fibre layers
to bend like a cantilever beam [32]. The uncut fibres either bend permanently or could sometimes return to their
original shape and position after the cutting tool progress forward due to the absence of plastic deformation.
Type II fibre buckling of the chip can be seen in Figure 16.a which occurs when the fibres are subjected to
compressive loading along their direction. Another very common phenomena were the irregular surface where
the fibre ends stick out of the surface with varying lengths as shown in Figure 16.b. The sticking out of the
fibres is because the cutting by fracture occurs at different points along the fibre lengths [32]. The inspection of
machined hole surface microstructure shows that fibres have partly impeded in the epoxy matrix which could be
due to the elastic recovery (see Figure 16.c). The fracture pattern of the matrix suggests it was stretched in Mode
I loading before fracture. The fibres on the machined surface are fractured perpendicular to their direction due to
the compression of the cutting tool edge against the surface and micro-buckling of fibres [32]. Usually, the
fibres lying at a relative orientation of 0° to the cutting edge sustains a minimum damage [41]. The 0° fibres
were sheared from the supporting epoxy matrix which led to the formation of the smooth fibre and socket
surfaces which are known as cusps as shown in Figure 17. Smearing of the softer composite matrix and
aluminium on the surface of the machined hole was present when drilling at high spindle speed and feed rates
due to the high cutting parameters. The smearing of the laminate constituents was more present near the exit of
the hole due to the rise of cutting temperatures with depth. SEM revealed that the region between fibre
interlayers tended to separate exclusively either at the fourth or the last glass fibre layer which indicates that it is
a possible location for internal delamination. The separation occurred when drilling at spindle speeds of
=9000 rpm and feed rates of = 900 mm/min in GLARE 3. Also at a spindle speed of =1000 rpm and
feed rates of = 600 and 900 mm/min in GLARE 3 and GLARE 2B 8/7 as shown in Figure 24. The separation
within fibre layers in GLARE 3 was more severe than in GLARE 2B. The delamination in the last glass fibre
layer is due to push-out delamination due to the action of the feed forces which pushes away the laminate
directly under the drill especially as it approaches the end of the workpiece.
Figure 17: Smooth fibres –Cusps- formed at 0° fibre orientation in GLARE 2B 11/10
The uncut thickness becomes smaller and the resistance to deformation decreases until the loading exerted by
the feed force exceeds the interlaminar bond strength leading to what is known as push-out delamination [52].
The delamination in the middle of GLARE 3 laminate is due to the continuous shear and compressive action of
the cutting tool while cutting through GLARE, which creates three-dimensional interlaminar stresses in glass
fibre layers that could lead to internal delamination between adjacent plies. The rotational and axial motion of
the cutting edge of the drill towards the bottom of the workpiece tends to pull the abraded material away along
the flute by introducing a peeling forces acting upwards against the downward thrust force, the material spirals
up before it is machined completely causing the upper laminas from the uncut portion to separate from the lower
Figure 18: Delamination in glass fibre layer in (a) GLARE 3 8/7 at = 9000
, = 900
/
(b)
GLARE 2B 8/7 at = 1000
, = 600
/
The different characteristics of metals and composites in GLARE can affect the machining performance and
result in GLARE. For example, the CT scan images revealed that some of the glass fibre plies were eroded in
the GLARE stack by metallic chips, the phenomenon which is known as composite erosion is caused by the
evacuation of metallic chips during drilling as shown in Figure 19Figure 18.a. The layered structure of GLARE
puts glass fibre layers between two metallic sheets can be compared to a composite workpiece stacked between
two metallic plates. Peel up and push-out delamination was minor in glass fibre layers which as mentioned
earlier by Tsao and Abrao that the use of a backup plate can significantly reduce the delamination by providing
support and minimising deformations [47, 48]. The erosion could be also attributed to the act of interlayer burrs
along with evacuated chips eroding the inner surface of the hole, this was observed in the work of Zitoune et al.
and Shyha et al. when drilling composite metal stacks [25, 31].
In GLARE, there are two different types of interfaces: the interface between aluminium sheets and the epoxy,
and the interface between the fibres and the epoxy. Both interfaces contribute in load transfer in the laminate
[2]. The bond between aluminium and epoxy and between the fibre-epoxy interfaces can be degraded by the rise
of temperatures in the laminate. This is evident by the erosion in glass fibre layers observed on the hole
boundaries as shown in Figure 19Figure 18.b. Surface erosion is common when drilling composites and could
be attributed to the fibre breakouts when the cutting direction is 45° to the fibre direction [53]. The softening of
the fibre-matrix during the drilling process reduces its ability to transfer the local strain disruptions to the fibres
and provide them with sufficient stability against the cutting action of the twist drill [41]. As a result, fibres
fracture and debonding can be seen on the edge of the hole when the feed force exceeds the ultimate strength of
the fibre causing them to fail in compression, tension or shear depending on the relative orientation of the fibre
with respect to the cutting edge of the drill [41]. The erosion can be seen to happen in ±45° (see Figure 19.b)
due to the engagement of the primary cutting edge into the workpiece [54]. The interaction of the cutting tool
with the workpiece causes compressive and bending loads on the fibres in that region eroding them by pushing
them down with respect to the transverse plane causing fibre pull outs.
Figure 19: Erosion of glass fibre layers in (a) GLARE 2B 11/10 at = 3000
= 300
/
(b)
GLARE 3 8/7 at = 3000
= 300
/
glass-epoxy matrix and those formed from aluminium sheets. In machining aluminium, the ductile material
flows plastically and is pushed ahead of the cutting tool until the compression stresses are high enough to cause
plastic deformation [32] and the chips undergo further deformations while chip flows upward exiting the
workpiece along the face of the cutting tool. In machining composites such as glass fibres, the cutting
mechanism takes place due to fracture of the matrix and fibres. The formed chip was collected after the drilling
of each hole and the analysis of the chip shows that fragmented and powdery chips were formed when cutting
through the glass fibre layers due to the brittle fracture of the fibres-matrix in the laminate [32]. The chips
collected from GLARE 2B and 3 were similar to same spindle speeds and feed rates, so only chips collected
from GLARE 2B are shown in Figure 20. Regarding the chips formed from aluminium sheets, it was observed
that the chip thickness and length increased with the increase of the feed rate and decreased with the increase of
parameters. For example, drilling at spindle speeds of = 1000 rpm, long type helical chips were formed at
feed rates of =100 mm/min as shown in (1). Increasing the feed rate to =300 mm/min produced short type
helical chips as it can be seen from (2). Increasing the feed rate further to =600 and 900 mm/min as in (3 and
4) produced loose fragmented chips with glass fiber layers attached to them which is an indication of excessive
feeding. Drilling at =3000 rpm and feed rates of =100 and 300 mm/min produced long conical helical
chips as shown in (5 and 6). Increasing the feed rate further to =600 and 900 mm/min produced snarled- long
helical chips and snarled conical helical chips, respectively as shown in (7 and 8). At spindle speed of =6000
rpm, long washer type helical chips were produced using feed rates of = 100 mm/min as shown in (9) and
washer type snarled helical chips at 300,600 and 900 mm/min as shown in (10,11,12). Finally drilling at spindle
speeds of =9000 rpm and feed rates of =100 mm/min produced snarled ribbon chips as shown in (13),
while at drilling under the feed rates of =300,600 and 900 formed conical chips with long ribbon from one
end. Generally, the chips formed at low feed rates of =100 mm/min at various spindle speeds were associated
with a good surface finish. In addition, chips produced at feed rates of =100 and 300 mm/min and spindle
speeds of =1000 and 3000 rpm were evacuated easily from the cutting zone and did not wrap around the
cutting tool. Long and continuous stringy chips with small curling were formed at high feed rates of =600 and
900 and high spindle speeds of = 6000 and 9000 rpm. They chips tended to wrap around the cutting tool
which would turn to be unproductive for continuous an automated machining operations where chip formation
due to drilling large number of holes should not stop the production [55]. These long and continuous chips were
found to damage to the borehole surface which clearly indicates a trade-off between spindle speeds and feed
rates in order to achieve good hole quality and efficient machining process.
4.0 Conclusions
The current work investigates the influence of cutting parameters (Feed rate and spindle speed) on the machined
hole quality when drilling GLARE fibre metal laminates. The quality of drilled holes was inspected in terms of
its burr formation, circularity error and hole size. The work also focuses on analysing the morphological
characteristics of the borehole surface via microstructural images taken by SEM and computerised tomography
(CT) scan to inspect for damage and delamination, the following can be concluded:
• The variation of hole size between top and bottom locations was greater in GLARE 3 than in GLARE
2B.
• Hole size is influenced by fibre orientation. Undersized holes are common when drilling GLARE fibre
metal laminates and hole size drilled in GLARE laminates is likely to shrink with depth.
• A reduction in hole size at exit can be seen with increasing the spindle speed and decreasing feed rate.
• Circularity error in GLARE 3 was greater than in GLARE 2B at the bottom location. Circularity error
increased with workpiece thickness and the increase in the feed rate.
• Different feed rates and spindle speeds parameters produced different burr heights, thickness and chips
formations. Burr formation cannot be eliminated completely, however rather minimised by proper
selection of cutting parameters. Results show that smallest burrs occur when feed rate is minimum.
• Within the tested range of feed rates and spindle speeds, and for both grades of GLARE (2B and 3), the
best surface quality was achieved at a feed rate of 100 mm/min and various spindle speeds, this cutting
condition gave low burr formation, circularity error and close to nominal hole size.
• The borehole surface damage is related to the workpiece thickness and fibre orientation.
• Overall, it is observed that the dimensional variations in hole increased with the increase of spindle
speed. Distortions to the hole may also occur due to the mismatch in thermal expansion coefficients
along and transverse to the fibre directions and between the epoxy matrix and the reinforcement fibres
which could lead to residual stresses and dimensional variations in the hole size and its circularity
error.
• Interlaminar failure during drilling is more likely to occur in GLARE 3 than in GLARE 2B because
laminates with cross-ply configuration are mechanically weaker than unidirectional ones which mean
that the influence of the feed rate is greater on GLARE 3 than on GLARE 2B.
• Various chip shapes and sizes were formed at different cutting speeds and feeds, chip thickness
Acknowledgments
The author would like to thank the Advanced Manufacturing Research Centre (AMRC) at The University of
Sheffield for the permission to use their facilities in this research project. The author would also like to thank Dr
Peter J. Kortbeek from FMLC and Professor Jose Sinke from DELFT University for their technical support on
machining GLARE and material supply. Special thanks to Mr Jacob Hawxwel from Sandvik-Sheffield for
conducting the CMM measurements. Last but not least, the first author would like to thank the Department of
Mechanical Engineering at The University of Sheffield for a PhD scholarship. Authors gratefully acknowledge
Professor Alma Hodzic for her support and guidance in this study.
References
[1] Norris G. Airbus A380: superjumbo of the 21st century: Zenith Imprint, 2005.
[2] Vlot A, Gunnink JW. Fibre metal laminates: an introduction: Springer, 2001.
[3] Graham-Cumming J. The Geek Atlas: 128 places where science and technology come alive: " O'Reilly
Media, Inc.", 2009.
[4] Mouritz A. Introduction to aerospace materials: Elsevier, 2012.
[5] Ostrower J. A Closer Look: Inside the wing of the A350 XWB. Flighglobal.com: flightglobal; 2010.
[6] J.F.W.Coesel. Drilling Of Fibre-Metal Laminates [Master Thesis]: Delft University of Technology, 1994.
[7] Tyczynski P, Lemanczyk J, Ostrowski R. Drilling of CFRP, GFRP, glare type composites. Aircraft
Engineering and Aerospace Technology. 2014;86:312-22.
[8] Giasin K, Ayvar-Soberanis S, Hodzic A. An experimental study on drilling of unidirectional GLARE fibre
metal laminates. Composite Structures. 2015;133:794-808.
[9] Pawar OA, Gaikhe YS, Tewari A, Sundaram R, Joshi SS. Analysis of hole quality in drilling GLARE fiber
metal laminates. Composite Structures. 2015;123:350-65.
[10] Giasin K, Ayvar-Soberanis S, Hodzic A. The effects of minimum quantity lubrication and cryogenic liquid
nitrogen cooling on drilled hole quality in GLARE fibre metal laminates. Materials & design. 2016;89:996-
1006.
[11] Giasin K, Hodzic A, Phadnis V, Ayvar-Soberanis S. Assessment of cutting forces and hole quality in
drilling Al2024 aluminium alloy: experimental and finite element study. The International Journal of Advanced
Manufacturing Technology. 2016:1-21.
[12] Giasin K, Ayvar-Soberanis S, Hodzic A. Evaluation of cryogenic cooling and minimum quantity
lubrication effects on machining GLARE laminates using design of experiments. Journal of Cleaner Production.
2016;135:533-48.
[13] Giasin K, Ayvar-Soberanis S, French T, Phadnis V. 3D Finite Element Modelling of Cutting Forces in
Drilling Fibre Metal Laminates and Experimental Hole Quality Analysis. Applied composite materials. 2016:1-
25.
[14] Giasin K, Ayvar-Soberanis S. Evaluation of Workpiece Temperature during Drilling of GLARE Fiber
Metal Laminates Using Infrared Techniques: Effect of Cutting Parameters, Fiber Orientation and Spray Mist
Application. Materials. 2016;9:622.
[15] Hocheng H, Tsao C. Comprehensive analysis of delamination in drilling of composite materials with
various drill bits. Journal of Materials Processing Technology. 2003;140:335-9.
[16] Amritkar A, Prakash C, Kulkarni AP. Development of temperature measurement setup for machining.
World Journal of Science and Technology. 2012;2:15-9.
[17] Gillespie L. Countersinking handbook: Industrial Press Inc., 2008.
[18] Stringer P, Byrne G, Ahearne E. Tool design for burr removal in drilling operations.
[19] PILNÝ BL. High speed drilling of aluminium plates [Master]: BRNO Univresity of technology, 2011.
[20] Gillespie LK. Deburring precision miniature parts. Precision Engineering. 1979;1:189-98.
[21] Choi IH, Kim JD. Electrochemical deburring system using electroplated CBN wheels. International Journal
of Machine Tools and Manufacture. 1998;38:29-40.
[22] Sureshkumar M, Lakshmanan D, Murugarajan A. Experimental investigation and mathematical modelling
of drilling on GFRP composites. Materials Research Innovations. 2014;18:S1-94-S1-7.
[23] Vankanti VK, Ganta V. Optimization of process parameters in drilling of GFRP composite using Taguchi
method. Journal of Materials Research and Technology. 2014;3:35-41.
[24] Krishnaraj V, Prabukarthi A, Ramanathan A, Elanghovan N, Senthil Kumar M, Zitoune R, et al.
Optimization of machining parameters at high speed drilling of carbon fiber reinforced plastic (CFRP)
laminates. Composites Part B: Engineering. 2012;43:1791-9.
[25] Zitoune R, Krishnaraj V, Collombet F. Study of drilling of composite material and aluminium stack.
Composite Structures. 2010;92:1246-55.
[26] Gowda BMU, Ravindra HV, Gurupavan HR, Ugrasen G, Prakash GVN. Optimization of Process
Parameters in Drilling Al-Si3N4 Metal Matrix Composites Material Using Taguchi Technique. Procedia
Materials Science. 2014;5:2207-14.
[27] Sreenivasulu R. Optimization of Burr size, Surface Roughness and Circularity Deviation during Drilling of
Al 6061 using Taguchi Design Method and Artificial Neural Network. Independent Journal of Management &
Production. 2015;6:093-108.
[28] Abdelhafeez AM, Soo SL, Aspinwall DK, Dowson A, Arnold D. Burr Formation and Hole Quality when
Drilling Titanium and Aluminium Alloys. Procedia CIRP. 2015;37:230-5.
[29] Shanmughasundaram P, Subramanian R. Study of parametric optimization of burr formation in step drilling
of eutectic Al–Si alloy–Gr composites. Journal of Materials Research and Technology. 2014;3:150-7.
[30] Hellstern C. Investigation of interlayer burr formation in the drilling of stacked aluminum sheets: Citeseer,
2009.
[31] Shyha I, Soo S, Aspinwall D, Bradley S, Perry R, Harden P, et al. Hole quality assessment following
drilling of metallic-composite stacks. International Journal of Machine Tools and Manufacture. 2011;51:569-78.
[32] Sheikh-Ahmad JY. Machining of polymer composites: Springer, 2009.
[33] Alderliesten RC. Fatigue crack propagation and delamination growth in Glare: DUP Science, 2005.
[34] Phadnis VA, Makhdum F, Roy A, Silberschmidt VV. Drilling in carbon/epoxy composites: Experimental
investigations and finite element implementation. Composites Part A: Applied Science and Manufacturing.
2013;47:41-51.
[35] Schäfer F. Entgraten : Theorie, Verfahren, Anlagen. [Mainz, Germany]: Krausskopf-Verlag, 1975.
[36] Bi S, Liang J. Experimental studies and optimization of process parameters for burrs in dry drilling of
stacked metal materials. The International Journal of Advanced Manufacturing Technology. 2011;53:867-76.
[37] Dornfeld D, Lee D-E. Machine design for precision manufacturing: Springer, 2008.
[38] Krulikowski A. Geometric Dimensioning and Tolerancing: Cengage Learning, 1998.
[39] De Vries TJ. Blunt and sharp notch behaviour of Glare laminates. 2001.
[40] Hagenbeek M. Characterisation of fibre metal laminates under thermomechanical loadings: TU Delft, Delft
University of Technology, 2005.
[41] KHAN ZM. A study of the drilling of advanced carbon fibre composites. University of Salford: University
of Salford, 1991.
[42] Velayudham A, Krishnamurthy R, Soundarapandian T. Evaluation of drilling characteristics of high
volume fraction fibre glass reinforced polymeric composite. International Journal of Machine Tools and
Manufacture. 2005;45:399-406.
[43] Vijayaraghavan L. Machining Of Composites An Overview. Journal on Design and Manufacturing
Technologies. 2007;1.
[44] Kurt M, Kaynak Y, Bagci E. Evaluation of drilled hole quality in Al 2024 alloy. The International Journal
of Advanced Manufacturing Technology. 2008;37:1051-60.
[45] coromant S. Machining carbon fibre materials. Sandvik coromant user’s guide - composite solutions 2010.
[46] Strong AB. Fundamentals of composites manufacturing: materials, methods and applications: Society of
Manufacturing Engineers, 2008.
[47] Tsao C, Hocheng H. Effects of exit back-up on delamination in drilling composite materials using a saw
drill and a core drill. International Journal of Machine Tools and Manufacture. 2005;45:1261-70.
[48] Abrate S, Walton D. Machining of composite materials. Part I: Traditional methods. Composites
Manufacturing. 1992;3:75-83.
[49] Karimia NZ, Heidary H, Kianfar P, Hasani M, Minak G. The Effect Of Support Plate On Drilling-Induced
Delamination. Acta Polytechnica CTU Proceedings. 2016;3.
[50] Chung DDL. Composite Materials: Science and Applications: Springer London, 2010.
[51] Park C, Mcmanus HL. Thermally induced damage in composite space structure: Predictive methodology
and experimental correlation. 1994. p. 26.
[52] Ho-Cheng H, Dharan C. Delamination during drilling in composite laminates. Journal of Engineering for
Industry(Transactions of the ASME). 1990;112:236-9.
[53] Mallick P. Composites engineering handbook: CRC, 1997.
[54] Merino-Pérez JL, Royer R, Merson E, Lockwood A, Ayvar-Soberanis S, Marshall MB. Influence of
workpiece constituents and cutting speed on the cutting forces developed in the conventional drilling of CFRP
composites. Composite Structures. 2016;140:621-9.
[55] Songmene V, Zaghbani I, Kouam J. Machining and machinability of aluminum alloys.