1-s2.0-S0007850615000426-main

age

ire. Inage,rst

ithteelwase inwasowingireeen5%.age

CIRP Annals - Manufacturing Technology xxx (2015) xxxxxx

, but

y. In

ith

bris

ions

sed.

IRP.

.

G Model

CIRP-1301; No. of Pages 4

Contents lists available at ScienceDirect

CIRP Annals - Manufacturing Technology

journal homepage: http: / /ees.elsevier.com/cirp/default .aspWire breakage and deection caused by nozzle jet ushing in wire EDM

A. Okada (2)a,*, T. Konishi a, Y. Okamoto a, H. Kurihara b

aGraduate School of Natural Science & Technology, Okayama University, Okayama 7008530, Japanb EDM R&D Division, Makino Milling Machine Co., Ltd., Aiko-gun 2430303, Japan

1. Introduction

In wire EDM, smooth exclusions of debris and bubbles from thegap and reduction of wire vibration are important to obtain a stablemachining performance [1,2]. Much debris stagnation in the gapand large wire vibration result in wire breakage, low removal rate,and low shape accuracy [36]. The debris exclusion is convention-ally done by jet ushing using nozzles. In general, high ow ratefrom the nozzles is more effective for smooth debris exclusion butit may bring the wire breakage [7,8].

For the wire vibration and breakage phenomena, Dauw [3,9]analyzed the wire deection due to discharge explosive forces. Han[4] simulated the rough cut surface geometry with vibrating wire.Rajurkar [10] investigated the inuence of sparking frequency onwire breakage to develop an on-line WEDM monitoring system.Mohri [11] and Obara [12] also investigated the wire vibration andthe model was proposed. However, they are mostly based on theeffects of discharge or wire conditions, and there is no paperfocusing on the quantitative effect of hydrodynamic force due to jetushing. In addition, frequent wire breakage at a particularmachined kerf length is well known in the practical wire EDM.However, ow elds around the wire, hydrodynamic stressdistributions on the wire due to jet ushing and their inuenceson the wire breakage have not yet been claried sufciently, sincesuch unsteady ow eld is not easy to clarify and a precise in-process observation of debris movements is difcult [13,14].

In this paper, the inuence of machined kerf length on the wirebreakage is experimentally investigated. Furthermore, ow eldsand debris residence time in the kerf, hydrodynamic stressdistributions acting on the wire, and the wire deections due tojet ushing are numerically analyzed by computational uiddynamics (CFD) and structural analysis. Based on the analyzed

results, optimum jet ushing conditions to prevent wire breakare discussed.

2. Inuence of machined kerf length on wire breakage

In the manufacturing site using a wire EDM, frequent wbreakage at a particular machined kerf length is well knownorder to clarify the effect of machined kerf length on wire breakwire breakage frequency with machined kerf length was investigated by the experiments using wire ED machine.

Fig. 1 shows distribution of wire breakage frequency wmachined kerf length when a straight kerf was machined into splate of 10.0 mm in thickness using a brass wire. Duty factor set to higher value than the machine-maker-recommended onorder to make the wire breakage easier to occur. Jet ushing applied using upper and lower nozzles, in which the jet direction was along with the wire running. The machinconditions are shown in Table 1. As shown in the gure, wbreakage frequently occurs when the kerf length is betw1.0 and 2.0 mm, and the frequency until 2.0 mm is about 8When longer kerf than 3.0 mm can be machined, the wire break

A R T I C L E I N F O

Keywords:

Wire EDM

Flow

Wire breakage

A B S T R A C T

High ow rate in nozzle jet ushing is effective for smooth debris exclusion from the wire EDM gap

this leads to large wire deection and vibration, resulting in the wire breakage and low shape accurac

this paper, the inuence of nozzle jet ushing on wire breakage was experimentally investigated w

varying the machined kerf length and machining conditions. Furthermore, the ow elds and de

residence time in the kerf, hydrodynamic stress distributions acting on the wire, and wire deect

were numerically analyzed. Based on the analyzed results, the causes of wire breakage were discus

2015 C

* Corresponding author.

E-mail address: [email protected] (A. Okada). Fig. 1. Wire breakage frequency distribution with machined kerf length

http://dx.doi.org/10.1016/j.cirp.2015.04.034

0007-8506/ 2015 CIRP.Please cite this article in press as: Okada A, et al. Wire breakage and deection caused by nozzle jet ushing in wire EDM. CIRP Annals -Manufacturing Technology (2015), http://dx.doi.org/10.1016/j.cirp.2015.04.034

neveworfreq

3. C

FEDMmoddeiocondelecregibouwerto 6mac0.5 mmac

Ta cosoftwof numto ththe The werturb

Acircupresthe surfassoforcsinc

difcult under the current CFD techniques. However, the verica-tion by the high-speed observation in our previous study provedthat the CFD analysis results could well simulate the actual owelds and debris movement in the gap without considering theseeffects [8,13].

4. Effect of machined kerf length on debris exclusion

Fig. 3 shows the analyzed ow elds in the kerf with varying themachined kerf length. When the machined kerf length is as short as

Table 1Machining conditions.

Workpiece SKD 11 (t = 10.0 mm)

Wire electrode Hard brass 200 mm in dia.Jet ushing nozzle diameter 6.0 mm

Wire running speed 10.0 m/min

Wire tension 12 N

Working uid Deionized water

Nozzle stand-off distance 0.5 mm

Flow rate from nozzles 6.0 L/min

Wire ED machine Sodick AQ 550 L

TableCFD

Jet

Wi

Flu

Flu

Noz

Flo

A. Okada et al. / CIRP Annals - Manufacturing Technology xxx (2015) xxxxxx2

G Model

CIRP-1301; No. of Pages 4

PleMar occurs. Under different ow rates of jet ushing, for differentkpiece thickness of 50 mm in thickness, similar wire breakageuency distributions with kerf length were obtained.

FD analysis model

ig. 2 shows CFD analysis model for solving the ow eld ined kerf using jet ushing nozzles. This three dimensionalel is based on an actual wire EDM for steel plate usingnized water under 1st cut conditions. The details of CFD modelitions are listed in Table 2. In order to calculate the wiretrode deection with jet ushing, the model includes the insideons of upper and lower nozzles. On the upper and lowerndary surfaces, ow inlet circles of 6.0 mm in diametere set for nozzle jet ushing, in which the ow rate was set.0 L/min. The workpiece thickness is 10.0 mm and the length ofhined kerf LM is varied. The nozzle stand-off distance is set tom. The wire electrode diameter is 200 mm and the width of

hined kerf is 250 mm. Then the gap is 25 mm.he simulations presented in this study are processed by usingmmercial software package of STAR-CCM+ Ver.5.02. Thisare operates by solving the governing differential equations

the ow physics including NavierStokes equations byerical means on a computational cell. The cell size adjacente wire electrode was small enough to simulate precisely, andother parts were a little coarse for saving computational time.uid ows, the debris tracks and the pressure distributionse calculated by a nite volume method as an unsteadyulent ow with Ke model [15].

downward velocity of 10 mm/min was given to the wiremference surface to realize the actual wire running. Asure boundary condition was set to a level of 10 mm aboveupper surface workpiece. No slip condition was applied to theaces of workpiece, nozzle and wire. The effects of impact forceciated with discharge sparks, bubbles behavior, electrostatice acting on wire electrode, and wire vibration were neglected,e CFD analysis that considers these factors is impossible or very

0.5 mm, the ow from nozzle drifted out of the machined kerfwithout owing into the kerf. In the case of 1.0 mm kerf length, theow from the jet ushing nozzle branches into inside and outsideof the machined kerf. When the machined kerf length is 2.0 mm,the ow owing into the machined kerf increases and stagnationarea generates behind the wire at the middle region in the kerf. Inthis range of the machined kerf length, it was conrmed that theow of the working uid was unsteady, in which the direction ofthe ow in machined kerf alternately changes upward anddownward. The pressure distribution in the kerf also changesunsteady. When the kerf length is longer than 3.0 mm, most owfrom the jet ushing nozzle ows into the machined kerf and theow eld becomes steady. From these results, it is considered thatdebris exclusion and wire behavior become unsteady due to greatturbulence in the machined kerf when the machined kerf length isshorter than 2.0 mm.

In order to clarify the effect of machined kerf length on thedebris exclusion, a particle tracking analysis by Lagrangian liquidsolid ow model was done. Fig. 4 shows the model, in which5 debris particles are arranged in front of wire in each layer, andthere are 10 layers every 1.0 mm along the wire direction.Therefore, 50 particles in total generates in the front gap every1.0 s in the model. One example of debris particle trackingsimulation is also shown. The particle color indicates the residencetime of the particle in the kerf. Due to the generation of triangular

Fig. 2. CFD analysis model.

2model conditions.

ushing nozzle diameter 6.0 mm

re running speed 10.0 m/min

id density (deionized water) 6.676 102 kg/m3id viscosity (deionized water) 8.887 104 Pa szle stand-off distance 0.5 mm

w rate from nozzles 6.0 L/minFig. 4. Debris tracking analysis by two phase ow CFD simulation.

Fig. 3. Difference in ow eld with machined kerf length.ase cite this article in press as: Okada A, et al. Wire breakage and deection caused by nozzle jet ushing in wire EDM. CIRP Annals -nufacturing Technology (2015), http://dx.doi.org/10.1016/j.cirp.2015.04.034

assumed to hard brass, and the wire is constrained in xydirections at the upper and lower wire ends, considering the actualwire support with wire guides. Wire tension is realized by givingvertical tensile load at the both ends of wire. The structuralanalysis conditions are listed in Table 3. The wire tension is xed to12.0 N in the following. The model is divided into sufciently smallcomputational cells to simulate precisely. The wire deections arecalculated by a nite volume method.

Fig. 7 shows the wire deection in upper half of the analyzed areawhen the kerf length is varied. The machining direction is leftward,

e to thekerfgthithowtion

didrent

areires ofit ising

theowkerfned

ia.)

A. Okada et al. / CIRP Annals - Manufacturing Technology xxx (2015) xxxxxx 3

G Model

CIRP-1301; No. of Pages 4stagnation area behind the wire at the middle region of the kerfshown above, some particles are trapped by the stagnation areaand remain in the kerf for a long time.

From the simulation results, the debris particle residence timein the kerf with the machined kerf length was calculated andplotted in Fig. 5. As shown in the gure, the average particleresidence time in the kerf is longer, when it is shorter than 1.5 mm.This is because the ow from nozzle drifts out of the machined kerfwithout owing into the kerf, as shown above. When the lengthbecomes 2.0 mm, the residence time drastically decreases, sincethe ow from the nozzle toward the kerf increases. When the kerflength is longer than 2.5 mm, the residence time increases with anincrease of kerf length.

5. Hydrodynamic force acting on wire caused by jet ushing

There are two types of forces acting on the wire with jet ushingof working uid. One is pressure acting perpendicular to the wiresurface, and another is shear stress acting parallel. Fig. 6 showspressure distributions on the wire surface around the upper edgeof workpiece using the normal nozzle. At the upper edge ofworkpiece, the pressure acting on the front surface of wire is higherthan that on the back one. Also around the lower edge ofworkpiece, the situation is the same. Furthermore, the pressureacting on the wire inside the kerf is lower, and the differencebetween pressures acting on the front and back wire surfaces isextremely small inside the kerf. These results show that the wire isforced backward concentratedly only around the upper and loweredges of workpiece by the jet ushing. Shear stress distributions onthe wire surface were also calculated but the value wasapproximately 1 kPa at a maximum, which is much smaller thanthe maximum pressure acting on the wire 225 kPa. Therefore,shear stress acting on the wire surface can be neglected in thesimulation for the wire deection discussed later. In order tosimulate the wire deection with jet ushing, the distributions ofpressure acting on the wire surface by jet ushing obtained by theCFD analysis were given to the wire surface in the structuralanalysis model. The structural analysis of wire was attempted by acommercial program of ANSYS Rev.14.0. The wire material is

and the horizontal scale of the wire displacement is amplied heremphasize the difference in wire deection. It is understood thatwire electrode is deected backward under any machined length, and the displacement takes a maximum when the kerf lenis around 2.0 mm. Fig. 8 shows the variation of wire deection wkerf length. When the kerf length is shorter than 2.0 mm, the elds were not steady, as shown above. Then, the wire deecvalue uctuated with times even when the machined kerf lengthnot change. Therefore, ve wire deection values at ve diffetimes were calculated in each kerf length, and the mean valuesplotted with an error bar in the graph. As shown in the graph, the wdeection and the variations are larger than those in the caseother kerf length because of its very unsteady ow eld when shorter than 2.0 mm. In this case, the most ow from the jet ushnozzles goes out from behind the workpiece without owing intokerf. The wire deection takes maximum at 2.0 mm, since the from the nozzle branches into insideand outsideof the machined and the ow led in the kerf was most unsteady. When the machi

Fig. 5. Variation of debris particle residence time in kerf with machined kerf length.

Fig. 6. Pressure distributions on the wire surface around the upper edge ofworkpiece using normal nozzle.

Table 3Structural analysis conditions.

Wire material Brass (200 mm dDensity 8.4 g/cm3

Youngs modulus, Poisson ratio 106 GPa, 0.35

Distance between wire guides 27.0 mm

Wire tension 12.0 N

Fig. 7. Difference in wire deection with machined kerf length.

Fig. 8. Variation of wire deection with machined kerf length.Please cite this article in press as: Okada A, et al. Wire breakage and deection caused by nozzle jet ushing in wire EDM. CIRP Annals -Manufacturing Technology (2015), http://dx.doi.org/10.1016/j.cirp.2015.04.034

kerfleng

Iof whydobseis asobvideeworlengwaswascam

Twiththe diffewhianalbackthe smabe juthe

6. In

FLMvariresidconsthengreathe wthe 2.0 mvibr

Fig. 9ush

Fig. 1width

A. Okada et al. / CIRP Annals - Manufac4

G Model

CIRP-1301; No. of Pages 4

PleMa length is longer, the wire deection gradually decreases with kerfth, since the ow eld in the kerf becomes stable.n order to verify the accuracy of CFD and structural simulationsire deection, an actual wire deection caused by the

rodynamic force with jet ushing was measured. Thervation model is shown in Fig. 9. The workpiece thickness

thick as 150 mm, and the wire tension is as low as 1.0 N toously observe actual wire deection. For observing a wirection, a small notch was rst machined at the middle ofkpiece, as shown in the gure. Next, the kerf of 10.0 mm inth was wire-EDMed under a higher wire tension, and the wire

stopped at the edge of the notch. Then, the movement of wire observed through the small notch by using a zoom videoera when the wire is subjected to jet ushing.wo photographs are the images inside the notch with andout jet ushing. In the photographs, the right side black area isshadow of wire. As can be seen from the photographs, therence in position between wire front lines is about 86 mm,le the wire deection simulated by the CFD and structuralysis under the same ushing condition is 105 mm. With theward wire deection, the ow eld and pressure eld aroundwire changes in the observation model, which would leads toll decrease in the wire deection. Considering this factor, it candged that the simulations in the present paper well expressesactual wire deection.

uence of nozzle jet ushing on wire electrode

ig. 10 shows the variation of kerf width Wk with the kerf lengthwhen the straight kerf of 10.0 mm was wire-EDMed. Theations of wire breakage frequency, wire deection D and debrisence time Tr were shown again in the graph. The kerf width istantly large when the kerf length is shorter than 2.0 mm and a little decrease with the length. At the longer kerf length, ittly increases. The variations of kerf width would result fromire behavior during wire EDM and the debris accumulation in

gap. In other words, when the kerf length is shorter thanm the width becomes wide due to large wire deection and

ation with long debris residence time in the gap. Between

the ow eld becomes stable, and wire deection and debrisstagnation decrease. Therefore, there is a good correlation betweenthe wire breakage frequency and the wire deection. In wire EDM,it is well known that the factors affecting the wire deection areimpact force associated with spark occurrence, and electrical staticforce between the wire and the workpiece during the application ofopen voltage. It is concluded from these results that hydrodynamicforce due to jet ushing with nozzle is also one of the dominantfactor causing the wire deection and wire breakage. Therefore, inorder to prevent wire breakage, it would be effective toappropriately adjust the jet ushing conditions, such as ow rateand the nozzle stand-off distance, particularly when the machinedkerf length is short in early machining stage. Furthermore, it wasconrmed that the wire breakage frequency in the early machiningstage could be greatly reduced by applying the optimum jetushing conditions with small wire deection calculated using thestructural and CFD simulation.

7. Conclusions

The inuence of nozzle jet ushing on wire breakage wasinvestigated, and the wire EDM experiments showed that wirebreakage often occurred at a particular short machined kerf length.The causes of wire breakage were discussed based on simulationresults of debris stagnation in the kerf and wire deections. Thewire deection due to hydrodynamic force with jet ushing can beaccurately simulated by the developed simulation model. The wiredeection due to jet ushing becomes larger and the debris easilyaccumulates in the gap at the particular short kerf length, whichcauses high frequency of wire breakage.

References

[1] Schumacher BM (1990) About the Role of Debris in the Gap During ElectricalDischarge Machining. Annals of the CIRP 39(1):197199.

[2] Okada A, Uno Y, Nakazawa M, Yamauchi Y (2010) Evaluations of SparkDistribution and Wire Vibration in Wire EDM by High-speed Observation.Annals of the CIRP 59(1):231234.

[3] Dauw DF, Sthioul H, Delpretti R, Tricario C (1989) Wire Analysis and Control forPrecision EDM Cutting. Annals of the CIRP 38(1):191194.

[4] Han F, Kunieda M, Sendai T, Imai Y (2002) High Precision Simulation of WEDMUsing Parametric Programming. Annals of the CIRP 51(1):165168.

[5] Cetin S, Okada A, Uno Y (2004) Effect of Debris Accumulation on MachiningSpeed in EDM. International Journal of Electrical Machining 9:914.

[6] Takeuchi H, Kunieda M (2007) Relation Between Debris Concentration andDischarge Gap Width in EDM Process. International Journal of Electrical Ma-chining 12:1722.

[7] Masuzawa T, Cui X, Taniguchi N (1992) Improved Jet Flushing for EDM. Annalsof the CIRP 41(1):239242.

[8] Fujimoto T, Okada A, Okamoto Y, Uno Y (2012) Optimization of NozzleFlushing Method for Smooth Debris Exclusion in Wire EDM. Key EngineeringMaterials 516:7378.

[9] Dauw DF, Beltrami I (1994) High-Precision Wire-EDM by Online Wire Posi-tioning Control. Annals of the CIRP 43(1):193197.

[10] Rajurkar KP, Wang WM (1991) On-Line Monitor and Control for Wire Breakagein WEDM. Annals of the CIRP 40(1):219222.

[11] Mohri N, Yamada H, Furutani K, Narikiyo T, Magara T (1998) System Identica-tion of Wire Electrical Discharge Machining. Annals of the CIRP 47(1):173176.

[12] Obara H, Abe M, Ohsumi T (1999) Control of Wire Breakage During Wire EDM.International Journal of Electrical Machining 4:5358.

[13] Okada A, Uno Y, Onoda S, Habib S (2009) Computational Fluid DynamicsAnalysis of Working Fluid Flow and Debris Movement in Wire EDMed Kerf.Annals of the CIRP 58(1):209212.

[14] Hinduja S, Kunieda M (2013) Modelling of ECM and EDM Processes. Annals ofthe CIRP 62(2):775797.

[15] Crowe C, Sommerfeld M, Tsuji Y (1998) Multiphase Flow with Droplets andParticles, CRC Press, Boca Raton.

. Observation of wire deection caused by hydrodynamic force with nozzle jeting.

0. Inuences of wire deection and debris residence time on machined kerf and wire breakage frequency.ase cite this article in press as: Okada A, et al. Wire breakage andnufacturing Technology (2015), http://dx.doi.org/10.1016/j.cirp2.0 and 3.0 mm, the kerf width decreases due to the decrease in thewire deection with short debris residence time. At longer length,it becomes wider due to large increase in debris residence time inthe kerf.

In addition, it can be understood that the ow eld, the debrisresidence time, and the wire deection directly inuence the wirebreakage. When the kerf length is 1.02.0 mm, the wire breakagefrequently occurs due to the unstable ow eld, large wiredeection, and debris accumulation. When the machined kerflength is 2.03.0 mm, the wire breakage frequency decreases, since

turing Technology xxx (2015) xxxxxx deection caused by nozzle jet ushing in wire EDM. CIRP Annals -.2015.04.034

Wire breakage and deflection caused by nozzle jet flushing in wire EDM1 Introduction2 Influence of machined kerf length on wire breakage3 CFD analysis model4 Effect of machined kerf length on debris exclusion5 Hydrodynamic force acting on wire caused by jet flushing6 Influence of nozzle jet flushing on wire electrode7 ConclusionsReferences