Journal of Materials Processing Tech.

255 (2018) 263–274

Contents lists available at ScienceDirect

Journal of Materials Processing Tech.

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

A study on EDM debris particle size and flushing mechanism for efficient T
debris removal in EDM-drilling of Inconel 718
⁎
M. Tanjilul , Afzaal Ahmed, A. Senthil Kumar, M. Rahman
Department of Mechanical Engineering, National University of Singapore, 117575, Singapore

A R T I C L E I N F O A B S T R A C T

Keywords: Electrical discharge machining (EDM) is a non-contact machining process that removes material through spark
EDM erosion and is often used for drilling holes in difficult-to-cut materials such as Nickel-based super alloys.
Particle size However, this process becomes slower with the increase of the hole depth. This is because the flushing pressure
Suction becomes inefficient for the effective evacuation of process debris leading to secondary discharges and resulting
Flushing
in increased machining time. This study presents an innovative simultaneous flushing and vacuum-assisted
Particles trajectory
Simulation
debris removal system, which facilitates better debris removal for deep-hole EDM drilling. Improvement in the
drilling time and better surface roughness have been achieved using the developed setup. To evaluate the
performance of the vacuum-assisted debris removal system, a novel computational fluid dynamics model is also
proposed in this study. Experimentally measured particle size data of full length scale act as a reference for the
proposed numerical model.The analysis presented in this study provides significant insight into sizes of the
debris particles in different machining conditions, in full length scale using an automated image processing
technique. The presented numerical model can be used to investigate the various factors influencing the removal
of debris from the machining zone. The experimental and numerical components of this research complement
each other in the design of the presented vacuum-assisted debris removal system.

1. Introduction movement flushing or vibration supported flushing.

Key advantages of the electrical discharge machining (EDM) process
are the absence of a cutting force and the flexibility for machining ir-
respective of the material’s hardness. So, EDM is a feasible solution for
machining nickel-based superalloys having high hardness and low
machinability. Drilling is one of the most important machining pro-
cesses, and around 50–70% of all production time is spent making holes
(Kuppan et al., 2008). Successful implementation of high aspect ratio
EDM drilling relies on efficient flushing application. With continuous
production of debris particles by the process, failure to evacuate debris
effectively leads to short-circuiting, which tend to create machining
instabilities (Ayesta et al., 2016); this, in turn, causes jumping off the
electrode back and forth to pump out the accumulated debris due to
servo feed control of the machine. Under severe conditions, the elec-
trode can no longer progress, and the drilling depth becomes limited
(Liao and Liang, 2016). Although the effect of flushing on the erosion
process has been intensely studied for decades, flushing remains one of
the most difficult parameters to apply correctly (Liao and Liang, 2016).
There are several methods to implement flushing, such as internal
flushing, external flushing, suction assisted flushing, different electrode
However, obtaining successful debris evacuation is still a challen-
ging task once the electrode has advanced to a particular depth, which
hinders EDM drilling application for high aspect ratio holes. Few re-
searchers have investigated EDM debris evacuation. Koyano et al.
(Koyano et al., 2016) investigated the influence of external hydrostatic
pressure in die sinking EDM and showed that the material removal per
pulse in consecutive pulse discharges slightly increases as the external
pressure slightly decreases. Ayesta et al. (Ayesta et al., 2016) conducted
an experimental study on debris evacuation during slot EDM and ana-
lyzed how particles move into the groove using a high-speed camera.
Experimental results showed that best debris removal was achieved at
the topmost electrode jump height with continuous flushing (Ayesta
et al., 2016). The researcher also investigated a vibration assisted
system. Yu et al. (Yu et al., 2009) presented a method in which pla-
netary movement of an electrode with enhancement from ultrasonic
wave provided an unevenly distributed gap for better debris evacua-
tion. Liao et al. (Liao and Liang, 2016) showed that the depth of the
hole could be increased by 75% when aluminum alloy is drilled using a
10-degree downward angle (inclined feed) electrode feed. Kunieda
et al. (Kunieda and Masuzawa, 1988) proposed an innovative concept

⁎
Corresponding author.
E-mail address: mpemti@nus.edu.sg (M. Tanjilul).

https://doi.org/10.1016/j.jmatprotec.2017.12.016
Received 21 March 2017; Received in revised form 8 December 2017; Accepted 11 December 2017
Available online 12 December 2017
0924-0136/ © 2017 Elsevier B.V. All rights reserved.
M. Tanjilul et al.
for debris removal by changing the direction of the main axis of the
EDM from vertical to horizontal. However, for large electrode lengths,
deflection of the electrode could be a serious issue and this set back its
application to deep holes.
Based on the aforementioned literature analysis it can easily be
understood that in deep hole EDM drilling, the accumulation of debris
in the cavity, which causes an increase in process time, still remains
unresolved.
In this paper a novel flushing system is proposed for EDM drilling
which can evacuate debris particles effectively, even from deeper
depth. This system is a simultaneous application of flushing through
electrode and vacuum-assisted suction on the side gap. A better ma-
chining rate and surface roughness were achieved by using the vacuum-
assisted debris removal system. The performance of the proposed
flushing system was further evaluated through a particle trajectory
model. Experimentally measured full length scale particle data were
introduced into the CFD (Computation Fluid Dynamics) model for
particle trajectory analysis. So far, only full-scale debris particles re-
search was conducted by Murray et al. (Murray et al., 2016) using
Scanning Electron Microscope (SEM), Transmission Electron Micro-
scope (TEM), and an image processing technique. However, due to the
complexity of their measuring technique the debris distribution mea-
surement may not have been very accurate. In this paper, an automated
image analysis method used for particle size measurement in full length
scale is used. Particle trajectory data from the CFD model comply with
the experimental observation.
264
Journal of Materials Processing Tech. 255 (2018) 263–274
Fig. 1. (a) Overview of the experimental setup, (b)
3D printed suction cup on top of the 150 mm work-
piece, (c) conductivity monitoring, and (d) con-
taminated EDM fluid trapped in a dielectric trap.
2. Experimental setup
Experiments were conducted on a Sodick K1C drilling machine.
Fig. 1(a) shows the layout of the suction-assisted debris removal system,
where a 3D printed suction cup (Fig. 1b) sits on top of the workpiece
and is connected to a dielectric trap. A vacuum pump is connected to
the dielectric trap. A resin trap, which is usually used for composites
manufacturing, reduces the risk of resin entry into the vacuum pump
and the central vacuum lines. The resin trap has been modified here to
be used as a dielectric trap. The dielectric trap ensures that con-
taminated dielectric after machining are trapped by it (Fig. 1d), and this
reduces the risk of damage to the vacuum pump through fluid entry.
The dielectric trap includes vacuum regulation for process control and a
vacuum pressure monitoring option. The dielectric trap is connected to
a dielectric tank. The connection between the dielectric trap and tank is
controlled using a pressure relief valve; the relief valve is designed to
open at a predetermined set pressure to prevent the dielectric trap from
overflowing. The dielectric trap also has an aluminum lid with a glass
view port which enables dielectric level checking at any time. The inlet
of the through electrode is connected with the dielectric tank for con-
tinuous flushing. Any spilt dielectric from the suction cup zone goes
directly to the machine tank, which is linked to the dielectric tank. A
conductivity meter (ProfiLine Conductivity Meter) was used to monitor
the conductivity of the through electrode flushing (Fig. 1c). This helps
to monitor any changes to the conductivity of the dielectric by debris. A
constant conductivity was maintained throughout the experiment. An
M. Tanjilul et al.
Fig. 2. Detailed schematic of Vacuum assisted debris removal system.
EDM fluid known as Vitol KS, offered by Sodick Pte Ltd, with a con-
ductivity of 2.15 ms/cm, was utilised in this study as the EDM fluid/
dielectric.
Fig. 2 below shows a detailed schematic of vacuum assisted debris
removal system for further clarification.
3. Experimental results
3.1. Particle size analysis
EDM drilling was performed at four different current settings for a
short duration (30 s). When deep holes are drilled, the particle build up
leads to secondary discharges, which can tamper the original size of the
debris particles. However, an ideal size range of the debris particles is
required for simulation model reference. Thus, to capture the particles
in an ideal scenario, where no particles build-up happens, the 30 s
machining time, which corresponds to low depth, was chosen for par-
ticle collection.The current profile was measured, as shown in Fig. 3,
using a Tektronix A621 probe. Previous studies showed that the pulse
on time significantly influences the MRR (Material Removal Rate) while
265
Journal of Materials Processing Tech. 255 (2018) 263–274
drilling holes in Inconel 718 (Kuppan et al., 2008). Thus, to achieve a
higher MRR, a 20 μs pulse-on time and a 6μs pulse-off time were se-
lected as the machining parameters. Brass was chosen as the electrode
material due to the fact that brass and Inconel 718 both have similar
densities (brass: 8.73 gm/cm3, Inconel 718: 8.22 gm/cm3), which made
the particle analysis easier.
Debris collectors were placed around the workpiece, as shown in
Fig.4(a), to ensure the collection of debris at full length scale.
Debris samples were processed using a Shimadzu laser diffraction
particle size analyzer to yield particle diameter data. A spherical par-
ticle can be described using a single number—the diameter—because
every dimension is identical. Non-spherical particles can be described
using multiple length and width measures. These descriptions provide
greater accuracy, but also greater complexity. Thus many techniques
make the useful and convenient assumption that every particle is
spherical. The analysis process involves essentially taking the physical
size data (i.e. laser diffraction particle size) as the particles flow ran-
domly through the laser beam, and determining the size of the sphere
that could produce the same result. Although this approach is simplistic
and not perfectly accurate, the shapes of particles generated by most
Fig. 3. Current profile at different current
settings for particle collection.
M. Tanjilul et al.
industrial processes are such that the spherical assumption does not
cause serious problems (Scientific, 2017).
Fig. 5 shows the measurement principle using the laser diffraction
particle analyzer. First, the flow system (dispersing bath and flow cell)
is filled with reverse osmosis (RO) treated water through the circulation
pump. A blank measurement was taken with the RO treated water
circulating through the flow cell. This analysis was followed by the
addition of sample solution into the dispersing bath. The addition of
266
Journal of Materials Processing Tech. 255 (2018) 263–274
Fig. 4. (a) Debris collector placed around the work-
piece, (b) collected debris.
sample was stopped when the light intensity stabilised. Sonication in
the dispersing bath was turned on for two minutes before the mea-
surement was taken. The sample was measured a few times to check
consistency; Figs. 6 and 7 show the measured debris size for all the
samples.
From the debris particle data in Figs. 6 and 7, it has been identified
that particle sizes change significantly with the current. This can be
explained based on the changes in material removal mechanism with
Fig. 5. Particle measurement principle (SALD-2300, 2017).
M. Tanjilul et al.
changes in discharge energy. According to Wong et al. (Wong et al.,
2003) when the energy level is lower, the vaporisation phenomenon
also causes the removal of material. So, it is assumed that with in-
creasing discharge energy there is a shift from vaporisation to melting.
Consequently, a change in the particle diameter from smaller to bigger
is observed. Nevertheless, there is a literature gap in the area of EDM
particle sizes in full length scale.
3.2. Machining time
Comparative experiments (with and without vacuum-assisted) on
the flushing system were conducted upto a depth of 150 mm in Inconel
718. Brass electrodes with two D shaped (kick pipe) flushing channel,
8 MPa flushing pressure, 500 rpm electrode rotational spped were used
for both cases. A 50 mbar vacuum pressure was applied on the outlet
side to create a suction effect for the vacuum-assisted flushing. A 16%
improvement in the machining time was observed with the application
of vacuum-assisted flushing for a 150 mm drill depth. Both the holes
measured 4.5 mm across the diameter. The material removal rate
(MRR) for normal and vacuum-assisted were 190.00 mm3/min and
232.5 mm3/min respectively. Improvement in the machining time can
be explained based on two scenarios. Firstly, due to the higher pressure
differential in the combined flushing system, better fluid flow in the gap
was achieved which facilitated debris evacuation. Consequently, there
267
Journal of Materials Processing Tech. 255 (2018) 263–274
Fig. 6. Particle distribution in full length scale.
were less secondary discharges due to debris accumulation. The second
explanation is based on the existing literature. In electrical discharge
machining, gas bubbles are generated in the working gap (Kunieda
et al., 2005). Koyano et al. (Koyano et al., 2015) showed that the ma-
terial removal per single pulse discharge increases as the external
pressure decreases. This is due to the fact that the pressure drop around
the bubble, causes bubble expansion, which promotes material re-
moval.
3.3. Surface roughness
Fig. 9 shows the surface roughness analysis of a vacuum-assisted
and a regular EDM drilled hole. Two 150 mm deep holes were sectioned
after machining using wire-EDM. A Taylor Hobson profilometer
(0.8 mm cutoff length) was used to evaluate the roughness and surface
profile of the holes. The scanning was carried out in the middle 90 mm
length section of the holes. The average surface roughnesses for the
vacuum-assisted and the conventional EDM drilled hole were 7.32 μm
and 9.43 μm, respectively. Evidently, better surface roughness for the
vacuum-assisted case was observed. This improvement can be attrib-
uted to the fact that, the evacuation of the process debris was more
effective in the vacuum-assisted drilling. Consequently, there was less
accumulation of the processed debris, which lead to fewer secondary
discharges. Secondary discharges are known to cause high process
Fig. 7. Exploded view of Fig. 6 for smaller diameter
distribution.
M. Tanjilul et al.
instability, reduced surface roughness, and low machining speed
(Ayesta et al., 2016).
3.4. Discharge condition
Discharge condition was studied for the regular discharge and for
vacuum-assisted discharge cases. At 100 mm depth, the current and the
voltage profiles for regular and vacuum-assisted cases were captured.
Fluctuations in the current profile was observed for both the cases.
Fig. 10 below shows the current profile, measured using an oscillo-
scope, for a single discharge at around 100 mm depth. For the vacuum-
assisted case, a slightly higher current was observed. This could be
explained by the changes/fluctuations in gap voltage. Machine current
was set at around 48 A and the gap (inter-electrode) voltage between
Fig. 9. Comparison of surface roughness.
268
Journal of Materials Processing Tech. 255 (2018) 263–274
Fig. 8. Comparative study of two flushing systems.
electrode and workpiece was set at 10 V. However, at around 100 mm
depth, average actual machining current was lower than the set value.
To understand the fluctuations in the current profile, specifically the
higher current profile in the vacuum-assisted case, a Tektronix
-DMM7510 digital sampling multitier was used to capture the voltage
over a short duration of 1 s. Fig. 11 below shows the experimental setup
for the discharge condition study, and Figs. 12 and 13show the voltage
profile over a duration of 1 s for the regular and the vacuum-assisted
drilling processes.
10 V was set as the reference voltage for the servo control. The servo
control causes the servo-drive system to advance in operation, to hold,
or to retract the electrode.
The servo control mechanism keeps the electrode-to-workpiece
machining voltage equal to 10 V DC reference voltage. Thus, at 10 V,
M. Tanjilul et al.
position machining occurs. If the voltage is less than 10 V, the servo
control will retract the electrode. So, for highly efficient machining, a
10 V gap voltage is expected over a long duration. A high retraction
frequency of the electrode causes loss in machining time and it often
happens due to debris particles building up in the machining zone.
From Figs. 12 and 13 it is clear that the retraction frequency of the
electrode is higher for the regular drilling conditions compared with the
vacuum-assisted drilling conditions. This is due to the superior debris
evacuation in the vacuum-assisted drilling process. Consequently, less
back and forth of the electrode was observed, and a more consistent gap
voltage was achieved, with vacuum-assist. The amount of energy con-
sumed during each pulse on time was constant. The total energy is a
function of current, voltage, and time. Since the time is also a constant,
only the pulse voltage and current parameters can vary. Due to the
higher rate of retractions of the electrode for the regular case, the
voltage was relatively higher. Consequently, the current profile for the
regular drilling was lower compared to the vacuum-assisted case.
4. Particle trajectory modelling
The influence of suction/reduced hydrostatic pressure in the outlet
of EDM drilling was further analysed through particle trajectory mod-
elling. Two distinct stages are involved in developing the particle path
model for EDM debris using the commercially available Fluent soft-
ware: (i) determining the mathematical equation of the particle dis-
tribution from experimental data; (ii) setting up the CFD model and
simulating the particles’ trajectories.
4.1. Particle distribution
For the particle trajectory modelling, experimentally measured
particle size distribution data was defined by fitting the size distribution
data into an analytical equation (Rosin-Rammler). In this approach, the
complete range of particle sizes was divided into a set of discrete size
269
Journal of Materials Processing Tech. 255 (2018) 263–274
Fig. 10. Current profile at same current
settings for (a) normal discharge and (b)
vacuum-assisted discharge.
ranges. Each range was defined by a single stream that is part of the
group. The Rosin-Rammler distribution function, based on the as-
sumption that an exponential relationship exists between the particle
diameter (d), and the volume fraction of the particle (Yd) with diameter
greater than d, Yd is
d n
Yd = e−( d * ) (4.1)
*
Fluent refers to the symbol d in Eq. 4.1 as the mean diameter and
the symbol n as the Spread Parameter. To determine these inputs, first
experimentally measured particle data for 48.8 A (Section 3.1) was
converted into volume fraction format (Fig. 14). Sub-micron particles
were ignored in the distribution model to avoid complexity in model-
ling. Only 8% of particles by volume were sub-micron. Thus, it was
assumed that ignoring sub-micron particles would not affect the tra-
jectory model.
Next, the values of d* and n were derived such that the data in
Fig. 14 fit Eq. 4.1. The value for d* is obtained by noting that this is the
value of d at which Yd = = 0.368. From Fig. 14, it is estimated that this
occurs for d =11.266 μm. The numerical value for n is given by
ln(−lnYd )
n = d
ln( d * ) (4.2)
By substituting the given data pairs for Yd and d/d* into this equa-
tion, values for n were found and then an average n value was calcu-
lated. Doing so yields an average value of n = 0.6695 for the example
data above. The resulting Rosin-Rammler curve is compared to the
experimental data in Fig. 15.
Values for Yd and n, as well as the diameter range of the data and the
total mass flow rate for the combined individual size ranges, were
passed to Fluent.
4.2. Mathematical basis of the simulation model
Fluent predicts the trajectory of discrete phase particles by
Fig. 11. Experimental setup for the dis-
charge condition study.
M. Tanjilul et al. Journal of Materials Processing Tech. 255 (2018) 263–274

Fig. 12. Voltage profile for regular discharge.

Fig. 13. Voltage profile for vacuum-assisted discharge.

Fig. 14. Cumulative size distribution of particles

based on volume fraction for 48.8 A current.

integrating the force balance on the particle, which is written in a 18μ CD R e

FD =
Lagrangian reference frame. This force balance equates the particle ρp dp2 24 (4.4)
inertia with the forces acting on the particle and can be written (for the
x direction in Cartesian coordinates) as (Fluen, 2017) Here, u is the fluid phase velocity, up is the particle velocity, μ is the
molecular viscosity of fluid, ρ is the fluid density, ρp is the density of the
dup gx (ρp −ρ) particle diameter. R e is the relative Reynolds number, which is defined
= FD (u− up) + + Fx
dt ρp (4.3) as

ρdp (up−u)
Where, Re ≡
FD (u− up) is the drag force per unit particle mass and μ (4.5)

270
M. Tanjilul et al. Journal of Materials Processing Tech. 255 (2018) 263–274

Fig. 15. Comparison of experimental size distribution

and Rosin-Rammler curve fit for 48.8 A particles.

Fig. 16. Fluent model for EDM debris trajectory.

Here, dp is diameter of the particle.

Table 1
The drag coefficient, CD, is given by
Boundary conditions. a2 a
CD = a1 + + 32
Boundary DPM BC Physical meaning Re Re (4.6)

Outlet Escape Particles are free to leave at this boundary Where a1, a2 , a3 are constants that apply for smooth spherical particles
Wall Reflect Particles rebound off of the pipe wall over several ranges of R e and which are given by Morsi and Alexander
(Morsi and Alexander, 1972).
Eq. 4.3 incorporates additional forces (Fx ). Additional force arises
due to a pressure gradient in the fluid

271
M. Tanjilul et al. Journal of Materials Processing Tech. 255 (2018) 263–274

Fig. 17. Trajectory of the particles based on dia-

meter (color) distribution.

Fig. 18. Extended zoom view of the particle trajec-

tories.

Table 2 ⎛ ρ ⎞ du
Number of particles. Fx = ⎜ ⎟ up
ρ dx (4.7)
⎝ p⎠
Outlet Total Particles Escaped Incomplete
Integration of Eq. 4.3 with time yields the velocity of the particle at
Vacuum assisted 21,840 21,840 0
Without vacuum 21,840 20,748 1092
each point along the trajectory, with the trajectory itself predicted by

272
M. Tanjilul et al.
dx p
= up
dt (4.8)
Equations similar to 4.3 and 4.8 are solved in each coordinate di-
rection to predict the trajectories of the discrete phase.
Assuming all forces remains constant for each small time interval,
the trajectory of the path can be rewritten in simplified form as
dup 1 derstood easily. Improvement of the machining time shown in Fig. 8
= (u− up)
dt τp (4.9)
Here, τp is the particle relaxation time.
Eqs. 4.8 and 4.9 are solved simultaneously to determine the velocity
and position of the particle at any given time (Fluen, 2017).
Fluent uses a trapezoidal scheme to integrate Eq. 4.9 (Fluen, 2017):
upn + 1−upn
1 * n+1
= (u −up )
Δt τ (4.10)
Where n represents the iteration number, Δt is small time duration and
1 n
u* = (u + un + 1)
2 (4.11)
aun + 1 = un + Δtupn . ∇un (4.12)
∇u is the change is velocity at time Δt .
4.3. CFD model set up
The trajectories of the particles were simulated based on the as-
sumption that the transport of the debris particles is carried out by the
flow field. To model the flow field area, the radial gap distance was
measured from the experimental overcut data (Section 3.2) for a 48.8 A
current. The gap distance between the electrode and workpiece frontal
area (interface between electrode and workpiece) was considered to be
200 μm. The efficacy of flushing was investigated in the numerical si-
mulation by dispersing the particles from the primary machining zone
(interface between electrode and workpiece). The total volume of
particles introduced into the gap per unit time was consistent with the
fraction of nominal material removal rate. Electrode and fluid domain
walls were set as no-slip boundaries, and the fluid regime was con-
sidered to constitute a single continuous 3D area. The meshing of the
fluid domain proved to be intricate on account of the fact that the
machining gap was quite small relative to the machining depth. The
sectioned hole in Fig. 9 shows that the radial gap distance was not
exactly the same throughout the hole. However, for simplicity, the fluid
domain diameter was considered to be equal throughout the entire
length of the model. The densities of brass and Inconel 718 are similar
and thus, the density of brass (brass: 8.73 gm/cm3, Inconel 718:
8.22 gm/cm3) was considered for the particle flow modelling. Fig. 16
shows the fluent model for EDM debris trajectory.
Water-based EDM fluid (80% water) was used for the experimental
study. Thus the properties of water were selected for the fluid proper-
ties for simplification. Interaction with the continuous phase was
chosen to ensure that, in addition to continuous water region influen-
cing the particles, the particles also influence the momentum/energy of
the water. Turbulent flow was considered due to the high Reynolds
number in the inlet. Outlet pressure was set differently for normal and
suction-assisted simulations. For regular drilling, the outlet pressure
was set to 101,325 (atmospheric pressure), and for suction assisted
drilling it was set to 5100 Pa (based on the experimental data of
51 mbar). Discrete Phase Modelling (DPM) conditions for the bound-
aries were set as per the table below (Table 1).
4.4. Result and discussion for CFD modelling
The trajectory of each particle is computed over a large number of
steps as it passes through the flow domain. Since we know the mass and
273
Journal of Materials Processing Tech. 255 (2018) 263–274
the surface area of each particle, at each step, the solver can compute
the balance of forces acting on it. The overall trajectory can be de-
termined by integrating over the flow domain. Fig. 17 shows the tra-
jectory of the particles for the vacuum-assisted case. Ten possible paths
for each particle were considered (Fig. 18). Table 2 shows the number
of particles at the outlet for both the cases. From the simulation data in
Table 2, the performance of the vacuum-assisted flushing can be un-
can be correlated with the particle distribution results in Table 2.
From the simulation results, it is understood that, without the va-
cuum, some particles were caught in a recirculation zone. Thus, they
were not able to escape, and this is reflected as incomplete in Table 2.
Thus, there were secondary discharges due to debris accumulation,
resulting in production time loss. This simulation result complies well
with the experimental observation. From the experimental section, it
has been observed that the electrode experiences high retraction fre-
quency without vacuum. This is due to the debris particles building up
in the machining zone. Better machining rate was observed with va-
cuum assist.
5. Conclusion
In this paper, a novel flushing mechanism, an analysis of EDM
particle sizes, and a full-length EDM particle trajectory model are
proposed. The experimental and numerical components of this research
complement each other in the design of a unique flushing system. Path
modelling provides comprehensive insights into phenomena related to
experimental observations. Based on the experimental and CFD ana-
lyses, the following conclusions are drawn:
Simultaneous flushing and vacuum-assisted debris removal system
can effectively remove debris particles from high drilling depth.
Improved surface roughness was observed in the case of suction-
assisted drilling.
Machining current significantly influences the size of the debris
particles, and particles size increases with increasing machining cur-
rent.
Particle trajectories in deep hole EDM drilling can be successfully
modelled by the proposed discrete phase modelling technique.
Optimal machining depth, flushing pressure, and suction pressure
can be calculated by introducing experimentally measured debris par-
ticle data into a CFD model.
The proposed full length particle trajectory model for EDM drilling
can be a used to analyze the flushing performance of different shaped
EDM drilling electrodes (e.g. slotted electrodes, multi-channel).
Acknowledgement
The authors would like to thank Ms Zinx Zheng (Shimadzu Asia
Pacific Ltd) for her contribution to particle analysis. Authors would also
like to thank Mr Siong Keat Cheong (Tektronix Southeast Asia Pte Ltd)
for his contribution to discharge condition study. A*Star Singapore
(Grant no: R265-000-534-305) and Singapore Institute of
Manufacturing Technology (SIMTech) (Grant no: R265-000-518-504)
supported the work described in this paper.
References
Kuppan, P., Rajadurai, A., Narayanan, S., 2008. Influence of EDM process parameters in
deep hole drilling of Inconel 718. Int. J. Adv. Manuf. Technol. 38 (1–2), 74–84.
Ayesta, I., et al., 2016. Experimental study on debris evacuation during slot EDMing.
Proc. CIRP 42, 6–11.
Liao, Y.S., Liang, H.W., 2016. Study of vibration assisted inclined feed micro-EDM dril-
ling. Proc. CIRP 42, 552–556.
Koyano, T., et al., 2016. Study on the effect of external hydrostatic pressure on electrical
discharge machining. Proc. CIRP 42, 46–50.
Yu, Z.Y., et al., 2009. High aspect ratio micro-hole drilling aided with ultrasonic vibration
and planetary movement of electrode by micro-EDM. CIRP Ann. Manuf. Technol. 58
(1), 213–216.
M. Tanjilul et al.
Kunieda, M., Masuzawa, T., 1988. A fundamental study on a horizontal EDM. CIRP Ann.
Manuf. Technol. 37 (1), 187–190.
Murray, J.W., et al., 2016. Physical and electrical characteristics of EDM debris. J. Mater.
Process. Technol. 229, 54–60.
Scientific, H., 2017. A guidebook to particle size analysis. Horiba Scientific Irvine, CA
92618 USA.
SALD-2300, S., 2017. Laser Diffraction Particle Size Analyzer. Japan.
Wong, Y.S., et al., 2003. Investigation of micro-EDM material removal characteristics
using single RC-pulse discharges. J. Mater. Process. Technol. 140 (1–3), 303–307.
274
Journal of Materials Processing Tech. 255 (2018) 263–274
Kunieda, M., et al., 2005. Advancing EDM through fundamental insight into the process.
CIRP Ann. Manuf. Technol. 54 (2), 64–87.
Koyano, T., et al., 2015. Influence of external hydrostatic pressure on machining char-
acteristics of electrical discharge machining. CIRP Ann. Manuf. Technol. 64 (1),
229–232.
A. Fluen , Fluent User Guide/ Manual 2017.
Morsi, S.A., Alexander, A.J., 1972. An investigation of particle trajectories in two-phase
flow systems. J. Fluid Mech. 55 (2), 193–208.