Here - the ESAFORM 2008 Conference - INSA de Lyon

Development of a friction modelling method in dry cutting of AISI 316Laustenitic stainless steelsC. Bonnet 1 , F.Valiorgue 1,2 , J. Rech 1 , J.M Bergheau 1 , P.Gilles 2 , C.Claudin 11 Ecole Nationale d’Ingénieurs de Saint Etienne (ENISE),Laboratoire de Tribologie et Dynamique des Systèmes (LTDS),UMR CNRS 5513, 58 rue Jean Parot42023 Saint-Étienne, France e-mail: cedric.bonnet@yahoo.frfrederic.valiorgue@enise.frjoel.rech@enise.fr2 AREVA NP,92084 Paris La Défense, FranceABSTRACT: This paper aims to identify a friction model able to describe the friction coefficient at theinterface between the tool, the chip and workpiece during the dry cutting of a AISI316L austenitic stainlesssteel with TiN coated carbide tools. A new tribometer has been designed in order to reach relevant values ofpressures, temperatures and sliding velocities. This set-up is based on a modified pin-on-ring system.Additionally a numerical model simulating the frictional test has been associated in order to identify localphenomena around the spherical pin, from the standard macroscopic data provided by the experimentalsystem. It has been shown that the friction coefficient is mainly dependant on the sliding velocity, whereas thepressure has a secondary importance. Finally a new friction model has been identified based on this localsliding velocity.Key words: Friction; tribology; numerical modelling; cutting; coating1 INTRODUCTIONIn the context of global competition, complementaryexperimental and numerical approaches allow todevelop high performance cutting processes.However, understanding of the interface phenomena(Fig. 1) occurring at the tool-chip contact (secondaryshear zone) and at the tool-workpiece contact(rubbing zone) have to be improved [1].WorkmaterialRubbingzoneVcfVchVc : Cutting speedf : feedPrimary shear zoneChipCutting toolVchip~Vc.fhFig. 1. Illustration of the various strategic zones in cutting.The Coulomb model, with a constant coefficientusually used to simulate the friction, is not relevantbecause it doesn’t take into account the temperatureand pressure influences. In the case of steelmachining, usual cutting conditions lead to severetribological conditions: high velocities (60-600 m.min -1 ), high temperatures (up to 1000°C),high pressures (up to 2 GPa) [2-3].The pin-on-disc system is the most widely knownset-up for tribological investigations. However, it isunable to simulate temperature and pressure foundduring cutting [4]. Moreover, during a cuttingoperation, the workmaterial is never more in contactwith the tool. Based on these statements, manyscientists [5] have work on open tribo-system. Theyproposed configurations in which a pin is placed justafter a cutting tool during the machining of a tube’sflat face in order to rub on a refreshed surface (Fig.2a and 2b). Friction conditions are well reproducedbut the set-up is very difficult to perform, expensiveand the duration limited. As a consequence, a newconfiguration of frictional tests has to be designed.The principle proposed by Hedenqvist et al. [6] is apromising configuration (Fig. 2c). This paperpresent a new experimental set-up based on hissystem in order to investigate the friction coefficient

occurring at tool-chip-workpiece interfaces duringthe high speed dry cutting of a AISI 316L stainlesssteel with TiN coated carbide tools.(a)Fig. 2. Various technologies of tribometer (a) Olsson’stribometer [5], (b) Zemzemi’s tribometer [5], (c) Hedenqvist’stribometer [6].(c)2 EXPERIMENTAL SET-UP2.1 Material descriptionA cylindrical bar, made of AISI 316L, is fixed ontoa lathe chuck, as illustrated in Figure 3.4-Pneumatic jack6-DynamometerForcesmeas.FtFn2-Cutting toolrefreshingthe surface5-Pin holder(b)3- Pin :Carbide +TiN coating7-Thermistor1-Workpiece :316LHeat Fluxmeas.Fig. 3. Tribometer designed for cutting applications.A pin made of cemented carbide with a TiN coatingand having a spherical geometry Ø 9 mm is pressedonto the cylindrical surface by means of a jack.Thanks to the helical trajectory of pin, the surfacecontact is continuously regenerated. The pin ismaintained by an instrumented pin-holder able toprovide data about the instantaneous heat fluxentering into the pin [7]. It is fixed onto adynamometer, in order to provide the apparentnormal and tangential force (macroscopic forces).After each friction test, a cutting tool refreshes thesurface and a belt finishing operation is performed inorder to obtain a very low initial surface roughness.2.2 Testing conditionsTesting conditions have to be chosen in accordancewith the frictional conditions estimated along thetool/chip/workpiece interfaces. For instance, themachining of a AISI 316L with a TiN coated carbidetool in dry cutting conditions is around Vc~120m.min -1 . Based on Figure 1, it is of evidence that themacroscopic sliding speed at the tool-workpieceinterface is almost equal to the cutting speed ~ 120m.min -1 . On the chip-tool interface, the averagesliding velocity depends on the compression ratio β,which is around 2. So at this interface, themacroscopic velocity is 2 times lower than thecutting speed 60 m.min -1 . As a consequence, thecharacterization of the frictional properties at thetool/chip/workpiece interface needs performingfriction tests in the following conditions: slidingvelocity: 60-120 m.min -1 , pressure: 1-2 GPa [2].2.3 Experimental resultsThe sliding velocity was the single modifiedparameter in this work. A normal force equal toabout 1000 N has been applied onto pins, which leadto an average pressure of 1.8 GPa.The first output data provided by this set-up is theratio between the tangential and the normal force.This coefficient can be defined as an apparentfriction coefficient Ftµapp= . (1)FThe evolution of µ app versus sliding velocity isplotted in Figure 4. This confirms that it is notrelevant to consider the friction coefficient asindependent from sliding speed. Over 120 m.min -1 , amodification of the frictional behavior is observed.nWorkmaterial : AISI 316 L steelPin : Carbide + TiN coatingSphere diameter : 9 mmFig. 4. Apparent friction coefficient.

The evolution of heat flux transmitted to the pin φ pinversus sliding velocity is plotted in Figure 5. Itappears that heat flow increases with the velocity,which is coherent with the fact that more energy isproduced and dissipated. The frictional behaviourmay be modified over a certain value (saturation).The Johnson-Cook parameters used have beenchosen thanks to the works of Umbrello et al. [8].Moreover our model considers the hypothesis fromBowden et al. [9] with: µ plast describes the plasticdeformation part of µ app and µ adh corresponds to theadhesive friction coefficient: µapp= µadh+ µ . (3)plastThe model uses thermo-dependent properties for theWC/Co substrate [6] and the AISI 316L workpiece:Workmaterial : AISI 316 L steelPin : Carbide + TiN coatingSphere diameter : 9 mmFig. 5. Heat flux measured into the pin.3 NUMERICAL POST-TRAITEMENTA finite element model, simulating the sameexperimental conditions of plowing and friction, hasbeen developed. Macroscopic experimental data µ appand φ pin are used to fit the numerical model for 3sliding velocities: 60-90-120 m.min -1 .3.1 Description of the modelVsVsTable1. AISI 316L workpiece propertiesParameter Temperature ValueSpecific heat 20 450(J.kg -1 .°C -1 ) 300 545500 570800 625Thermal Conductivity(W.m -1 .°C -1 )1100 67020 14300 18500 21800 241100 29Density (kg.m -3 ) 20 8000300 7890500 7800800 76601100 75103.2 Thermal boundaries conditionsThe model includes the heat generation induced bythe friction in addition to the heat dissipated by theplastic deformation (plowing) of the workmaterial.As shown by Figure 7, the frictional heat flux φ frictionis dissipated at the interface in a narrow layer,whereas the plastic deformation heat flux φ plast isdissipated only in the workpiece in a large volume.Fig. 6 Geometry of the numerical friction model.The 3D model (Fig.6) is based on the work ofZemzemi et Al. [5]. Considering the high strain andstrain rate, an explicit formulation has been chosen.A thermo-mechanical model has been programmedwith ABAQUS explicit.A Johnson-Cook flow stress model has been used tomodel the AISI 316L steel. This model is dependenton strainε p, strain-rate ε&pand temperature T so asto observe softening phenomenon:σeqm⎡ ⎛ & ε ⎞⎤⎡np ⎛ T − T ⎞ ⎤0= ⎣⎡ A + B( εp) ⎦⎤ . ⎢1 + C.ln . ⎢1−⎥⎜ ⎥ε ⎟ ⎜ ⎟⎢⎣ ⎝&0 ⎠⎥⎦ ⎢ ⎝ TF− T⎣0 ⎠ ⎥⎦(2)Plastic deform ationheat fluxTransm itted to part 1Plastic deform ationheat fluxD iffused in part 2Friction heat fluxtransm itted to part 1Friction heat fluxtransm itted to part 2Fig. 7. Heat flux sources during friction tests.3.3 Fitting methodFor each testing condition simulated, the indentationdepth h, the adhesive friction coefficient µ adh and theheat partition coefficient α have to be determined. Inthis aim, an iterative method, presented in Figure 8has been developed. The experiments has provided

Investigation of friction in warm forging of AA6082B. Buchner, A. Weber, B. BuchmayrChair of Metal Forming, University of Leoben – Franz-Josef-Strasse 18, 8700 Leoben, AustriaURL: www.metalforming.at e-mail: metalforming@unileoben.ac.atABSTRACT: This paper presents an experimental investigation of friction in hot forging of AA6082, whichis a standard forging alloy in automotive engineering, mechanical engineering and in naval architecture, byemploying a modified ring-on-disc test. The experiments were performed with three different commercialgraphite-based lubricants and at various loads, sliding velocities and specimen surface conditions. In addition,some tests were performed without lubrication.KEYWORDS: Friction, Warm Forging, Aluminium1 INTRODUCTIONForged parts made of aluminium play a significantrole as components of light weight structures in theautomotive and aerospace industry. The main parametersinfluencing forging processes are the flow curvesof the specimen material, the heat transfer at the contactarea and the friction in the die-workpiece interface[1, 2]. The exact knowledge of the latter is paidmuch attention to as it affects power requirements,material flow, die filling, tool life and workpiece quality.In order to understand the tribological processesand interactions in the tool-workpiece interface, theinfluences of the load collective and the surface conditionson friction were investigated systematically. Theexperiments were performed with a facility based onthe ring-on-disc test introduced by Schey [3].2 EXPERIMENTAL WORKforce is realised by another load cell, the rotationalspeed of the specimen is calculated from the speed ofthe servo motor. The specimen is brought to forgingtemperature by an inductive heating system, whereasthe tool is heated by a heating sleeve. The lubricantis sprayed onto the tool by an automatic applicationsystem that allows a reproducible dosing. The testingdevice is controlled by a programmable logic controlunit (PLC), the data acquisition is realised by a measurementamplifier that is connected to a commercialpersonal computer.The tribometer has a maximum upsetting force of approximately100 kN and provides a maximum torqueof more then 800 Nm on the rotary disc. The rotationalspeed can be up to 1.7 s −1 . The sliding distanceis not constrained by the testing device. Themaximum temperature of the specimen is 1200 °C,the maximum temperature of the tool is 450 °C. Thetemperatures can be kept constant in an interval of±2.5 °C.2.1 Testing deviceFigure 1 shows the Rotational Forging Tribometer.The circular motion is supplied from the bottom sideby means of a servo motor and a bevel gear system.The compression force is applied by a hydraulic cylinderfrom the top side. The specimen (workpiece)is mounted on a rotary disc and transmits the frictionaltorque to the pivot-mounted ring-shaped tool.The tool is supported by a load cell via a lever armwhich enables an accurate measurement of the frictionaltorque. The acquisition of the compressionFigure 1: Rotational Forging Tribometer.1

2.2 Toolkit and specimen geometryFigure 2 shows the toolkit used in the investigations.The ring-shaped workpiece is placed in the cavity ofa container and compressed by an annular tool withthe same inner and outer diameter as the specimen.Sliding on the bottom face of the workpiece is preventedby preparing the bottom of the cavity with radialridges, and the container is split in order to alloweasy removal of the tested specimen. The temperaturesof die and workpiece are measured via thermocouples. Tool and specimen are made of hot worktool steel 1.2344 (hardened to 55 HRC) and AA6082,respectively.seven tests were performed with exception of series9, where just two experiments were carried out perload level due to heavy wear at some stages.The experiments were carried out in the followingway: First, the tool was brought to operating temperature.When the testing sequence was started bythe PLC, the specimen was heated, and the final temperaturewas kept constant for 180 s in order to allowtemperature equalisation. Then, the lubricant was appliedby the automatic spraying device (series 1–8).After lubricating, the execution of the test itself wasstarted. In order to avoid an interaction of differentinfluences, the specimens were compressed in a firststep and thereafter the rotation began.Table 1: Lubricant test matrix.lubricant typerec. dilution ratioA dispersion of graphite in water 1:15B dispersion of graphite in water 1: 7C emulsion of a dispersion of 2: 1graphite in water and mineral oilFigure 2: Toolkit used in the investigation.2.3 Experimental detailsThe experiments were performed with three differentcommercial graphite-based lubricants (see Table 1)and at various loads (20–150 MPa), sliding velocities(10–100 mm/s) and specimen surface conditions(turned, sand blasted). In addition, some tests wereperformed without lubrication. The changing parametersof the test series are summarised in Table 2.On the one hand, the sliding velocities were varied atthe same interface conditions (series 1–4), and on theother hand, the interface conditions were changed atconstant sliding speeds (series 5–9). Tool and workpiecetemperature were set to 250 °C and 450 °C, respectively,and the relative displacement was 70 mm.For each parameter set and when pick-up occurred,the tool was prepared with grit 800 sandpaper and3 µm polishing suspension. Within a parameter set,Table 2: Parameters of the test series.series no. lubricant surface cond. velocity, mm/s1 B sand-blasted 102 B sand-blasted 403 B sand-blasted 704 B sand-blasted 1005 B, 80 % 1 sand-blasted 406 B, 80 % 1 turned 407 A sand-blasted 408 C sand-blasted 409 without sand-blasted 401 In series 5–6, 20 % less lubricant than in series 1–4 was used.2.4 Evaluation of the testsThe evaluation of the experiments was performed inthree steps (see Figure 3):1. First, the stationary region of the experimentwas determined from the velocity curve: Constantconditions were assumed in the intervalwhere the actual velocity at time step i wasequal to or greater then the reference velocity.2. In the stationary region, the mean values of normalpressure σ n , friction stress τ f and frictioncoefficient µ were calculated by the following2

equations:σ n =τ f =t1f · (te)e −t b∑ σ n,i , (1)i = t bt1f · (te)e −t b∑ τ f ,i , (2)i = t bµ = τ fσ n. (3)Herein, t b and t e indicate the begin and the endof the stationary region and f is the samplingrate of data acquisition. σ n,i and τ f ,i are thenormal pressure and friction stress at time incrementi, respectively.only be explained by the adhesion theory, extensivewelding in the tool-workpiece interface has to be assumedin these cases. This assumption was verifiedby a visual inspection of the surfaces and by a comparisonof the lifting forces of the friction facility afterthe tests.The shear yield stress of AA6082 at an interface temperatureof 350 °C was determined to be k = 45–50 MPa. From the performed experiments, shearingof the specimen (and therefore sticking friction) canbe presumed at normal pressures equal to or greaterthan 65 MPa. Micrographs confirmed that (most of)the relative motion was done by shearing in subsurfacelayers of the specimens at all pressure levels.3. For easier illustration, the σ n -, τ f - and µ-values of each parameter set were averaged tothe mean values σ n,m , τ f ,m and µ m and thestandard deviations were calculated. In orderto avoid falsified results, apparent outliers werenot considered in this calculation.Figure 4: Results of the tests without lubrication.3.2 Results of the tests with lubricant3 RESULTSFigure 3: Evaluation of the measurements.3.1 Results of the tests without lubricantThe friction stresses and friction coefficients obtainedfrom the experiments without lubrication are presentedin Figure 4. Even at the lowest load level,the asymptotic behavior of the friction stress is wellpronounced; at normal loads higher than 65 MPa, amore or less steady state is reached. When regardingthe friction coefficient, a decrease from an initialvalue of µ = 1.83 (!) to a final value of µ = 0.34 isobserved. As a friction coefficient higher than 1 canFigure 5 presents the friction stress at different interfaceconditions. For comparison, the region of thevelocity dependent curves is indicated hatched. Exceptof series 5, all series have a maximum value at anormal stress of 130 MPa and have similar or lowervalues at contact loads of 150 MPa. However, thehigh friction stress of series 5 at a normal pressureof 150 MPa was due to material extruded in the gapbetween tool and container and did not indicate anothertendency. In contrast to the other curves, series 7shows a sharp rise from σ n = 100 to σ n = 130 MPa.When comparing the sand-blasted with the turned surface,the curve of the untreated specimen rises significantlysteeper than that of the sand-blasted workpiece.The lowest friction stresses are obtained with lubricantC (series 8), whereas the highest friction stressesare caused by the conditions present in series 7 (lubricantA). Figure 6 shows the friction coefficients independence of the surface conditions.3

In general, the friction coefficient increases significantlyat low normal pressures and stays approximatelyconstant or shows an light decrease at elevatedcontact stresses. Lubricant A has a well pronouncedlocal maximum at σ n = 130 MPa, and thefriction coefficient rises at high normal pressured inseries 5. However, the behavior of series 6 is contrary:Here, the friction coefficient rises from the beginning,reaches a maximum at a normal pressure of 130 MPa,and decreases slightly at σ n = 150 MPa.confirmed that shearing was restricted to the graphitelayer and the asperities; the bulk material of the specimenwas not affected.4 CONCLUSIONSThe present analysis aimed in the characterisation ofthe load collective and the interface conditions onfriction. From this point of view, especially two conclusionscan be made:1. As in solid lubricated interfaces (nearly) allshearing is done in the lubricant layer, the lubricantitself and the surface conditions of thefriction partners are the dominating parametersof the tribological system.2. When regarding the load collective, the effectof normal pressure was in the observed rangemore significant than the influence of the slidingvelocity.Figure 5: Friction stresses in the tests with lubrication. Theregion of the velocity dependent curves is indicated hatched.The results were compared to the results of a previousstudy [4] employing a pin-on-disc test for lubricantevaluation, and good qualitative agreement was foundin terms of friction coefficient and friction evolutionduring the tests.ACKNOWLEDGEMENTThe authors want to thank the federal province of Styria(”Zukunftsfonds Steiermark”, Project 19) for financing theproject and Fuchs Schmiermittel GmbH and Acheson Industriesfor the provision of the lubricants.REFERENCESFigure 6: Friction coefficients in the tests with lubrication. Theregion of the velocity dependent curves is indicated hatched.It has to be stated that all lubricants investigated reducedfriction significantly compared to the dry frictioncondition. In fact, lubricant A (series 7) thatshowed the poorest lubricating effect reduced the frictionstress about 80 %, and lubricant C (series 8) reducedfriction about more than 90 %. Moreover, alllubricants prevented the tool from wear. Micrographs[1] P. Groche and U. Weiss. Numerical identification of forgingparameters. In Modelling of Metal Forming Processes: Proceedingsof the Euromech 233 Colloquium, pages 237–244,Sophia Antipolis (France), 1988.[2] R. G. Snape, S. E. Clift, A. N. Bramley, and A. N.McGilvray. Forging modelling – sensitivity to input parametersusing FEA. In 12th National Conference on ManufacturingResearch – Advances in Manufacturing TechnologyX, pages 51–55, Bath (UK), 1996.[3] J. A. Schey. Tribology in Metalforming. Friction, Lubricationand Wear. American Society for Metals, Ohio (USA),1983.[4] B. Buchner, G. Maderthoner, and B. Buchmayr. Characterisationof different lubricants concerning the friction coefficientin forging of AA2618. Journal of Materials ProcessingTechnology, 198(1–3):41–47, 2008.4

Process parameters influence on friction coefficient in sheet formingoperationsE. Ceretti 1 , A. Fiorentino 1 , C. Giardini 21 University of Brescia, Dept. of Mechanical and Industrial Engineering - Via Branze 38, 25123 Brescia, ItalyURL: www.ing.unibs.it/tecmece-mail: elisabetta.ceretti@unibs.it; antonio.fiorentino@unibs.it2 University of Bergamo, Dept. of Design & Technologies - Viale Marconi 5, 20044 Dalmine (BG), Italye-mail: claudio.giardini@unibg.itABSTRACT: In conventional sheet forming processes, such as stamping or drawing, significant contactphenomena take place between workpiece and die surfaces. Especially, relative motion and normal loadsgenerate friction which influences some aspects of processes such as material flow, tools wear and life andtotal force needed to complete the process. In the current paper an experimental test campaign has beencarried out using a large scale pin-on-disk device designed and realized by the Authors to investigate theinfluence of pressure, sliding velocity and temperature. The purpose is to test the developed device and to findwhich and how these parameters mostly affect friction. The pin-on-disk test consists of two specimens, a pinand a plate representing respectively die and workpiece, which are compressed by means of a known forceand then moved one over the other. Compression and friction forces are sampled during the tests and thefriction coefficient is estimated as the ratio of these two forces. The tested materials are H13 die steel onFeP04 and AZ31 sheets.Keywords: sheet forming, friction, contact problems, temperature, pressure, velocity.1 INTRODUCTIONIn conventional manufacturing operation significantcontact phenomena take place between tool andworkpiece. Due to this contact and the resultingsurface interactions, friction is generated. Frictionknowledge is very important because it affects manyaspects of manufacturing processes, such as therequired force, the die wear and in some cases theprocess feasibility.To investigate and to determine the friction valueseveral tests, such as ring test [1] and strip test [2],and the Pin-on-Disk [3] have been introduced. Inrecent years many Authors have proposed newmethods to evaluate friction, for example themodified LDH [4], the Twist Compression Test [5].Among these tests Pin-on-Disk (PoD) allows todirectly measure the compression (F) and friction(T) forces and to estimate the friction coefficient(µ=T/F) using one single couple of surfaces incontact.In previous works the Authors studied the influenceof contact pressure, sliding velocity and materialroughness on friction coefficient by using a selfdesignedPin-on-Disk equipment [6, 7, 8]. In theseworks, the H13 steel and different die coatings weretested on AISI 5115 and on Series 1 Aluminiumpainted sheets. The idea was to identify thedependence of friction on the process parametersand to implement it in FE codes.The aim of the present work is to study the influenceof contact pressure, sliding velocity and temperatureon friction while forming different materials, namelydeep drawing steel FeP04 and AZ31 Magnesiumalloy sheets. Since this latter is characterized by ahigh formability in warm forming [9], thetemperature influence was studied too.2 EXPERIMENTAL EQUIPMENTThe experimental device (Fig. 1 left) is a large scalePin-on-Disk tribometer designed by the Authors [7].The friction coefficient is evaluated by pushing intocontact two components called Pin and Plate. ThePin (a cylinder with a diameter of 12 mm and acorner radius of 2 mm) stands for the die (or more ingeneral the tool) and is mounted on a carriage whichcan move orthogonally with respect to the platesurface. The Plate (flat) represents the workpieceand it is mounted on a disk that rotates so

determining a relative movement between the twosurfaces whose parallelism is guaranteed by theaccuracy of the pin realization and of its holdingsystem. The two bodies are kept in contact by meansof a hydraulic cylinder acting on the carriage. Theuse of a pressure transducer and load cells allows tomeasure the normal (F) and the friction (T) forcesacting between pin and plate. The dynamic frictioncoefficient can be estimated by using the Coulombrelationship (µ=T/F).The test data are collected by an acquisition systemsampling and processing the transducer signals.To carry out tests at different working temperaturesan additional device for heating the plate was used.This heating system (Fig. 1 right) consists of a baseplaced between the disk and the plate and containinga heating rod. A feedback system based onthermocouples positioned one under the plate andone in the rod, keeps the plate at the desired constanttemperature (up to 310°C).3 EXPERIMENTAL TESTSTests were conduced using H13 steel pins on AZ31magnesium alloy and on FeP04 deep drawing steelplates.The experimental campaign was designed as ageneral full factorial experiment with multiplefactors and multiple levels. Each factor combinationwas performed with 3 replications. The consideredfactors were: velocity and contact pressure for bothFeP04 and AZ31. Temperature influence wasstudied on AZ31 due to its higher formability atwarm temperature.Since the forces exchanged between punch, die andblank holder with the material under deformation arestrictly related with its yield strength in stampingoperations, the contact pressures were chosen asgiven percentage of the yield stress itself. In such away, it was also possible to consider the yield stressreduction due to the temperature influence for AZ31alloy. All tests were performed in dry conditions andwith the same sliding length.The tests summary is reported in Table 1.Table 1 – Design of the experiments.Plate FeP04 deep drawing steel sheet (t=2 mm)Pin H13 steelVelocity 21 – 42 mm/sPressure 7.5 – 15 – 27% σ 0Temperature 20°Cσ 0 170 MPaLubrication DryPlate AZ31 magnesium alloy sheet (t=1 mm) Pin H13 steelVelocity 21 – 42 mm/sPressure 7.5 – 15% σ 0Temperature 200 – 250 – 300 °Cσ 0 (T) 127 – 86 – 63 MPaLubrication DryThe working procedure adopted for the tests was:• specimens surfaces cleaning;• machine set-up and compression load (F)application;• temperature assessment (if necessary);• plate movement and data acquisition;• load removal.Each repetition was performed with a new pin andplate couple.4 EXPERIMENTAL RESULTS AND ANALYSISThe results were statistically analyzed and comparedto identify which of the tested parameterssignificantly affects the µ behaviour and how itdoes. Single and multiple interactions wereconsidered too.4.1 FeP04 resultsThe typical friction coefficient behaviour duringFeP04 tests is shown in Fig. 2. It is characterized byan initial transient (A) followed by the steady state(B).The friction coefficient for each test was estimatedas the average value in the steady state interval (B).MOTORDISKPLATETPINFPTHEATING RODANDTHERMOCOUPLECARRIAGECASELCHYDRAULICCYLINDERTHERMOCOUPLEPLATEFig. 1. PoD (left) and heating device (right).

Table 2 reports the experimental friction valueresults in terms of mean and standard deviation,estimated considering three repetitions for each testparameters combination.µ0.080.060.040.02FeP04 vs vs H13 H13 | p | p = = 7.5%0σ- 0 v - v = = 4242mm/sA00 0.5 1 time [s]Fig. 2. Example of detected friction profile for FeP04.Bpressure causes an increase of friction for both thesliding velocities. Considering sliding velocity Fig. 4seems to demonstrate a dependence of friction onvelocity, but this is not confirmed by the statisticaltest (see Fig. 3).4.2 AZ31 resultsA typical friction profile for AZ31 tests is shown inFig. 5: after an initial transient (A) the curve growsas the two surfaces slide (B). The slope in B intervalincreases as the temperature increases.Also for AZ31 sheets, the friction coefficient wasestimated as the mean value in the B interval.The results are reported in Table 3 and Fig. 6 wherethe parameters effect on friction is shown.Table 2 – FeP04 experimental results summary.Plate FeP04Pin H13Velocity Pressure ratio µ[mm/s] [% σ 0 ] mean σ7.5 % 0.080 0.01121 15 % 0.125 0.01827 % 0.143 0.0377.5 % 0.082 0.01742 15 % 0.094 0.00927 % 0.140 0.019pp*vvPareto Chart of the Standardized Effects(α=0.05)Standardized EffectFig. 3. Pareto Chart for FeP04 results.[% σ 0 ] p7.515.0 p27.0Interaction Plot (data means) for µ21 [mm/s]Fo limitFo42 [mm/s]0.140.120.10AZ31 vs H13 | T = 300°C - p = 15% σFeP04 vs H13 | p = 7.5% ∠ 0 - v = 0 - v = 21mm/s42 mm/sµAB0.40.30.20.100 0.5 1 1.5 2 2.5 time 3[s]Fig. 5. Example of friction profile for AZ31.Table 3 – AZ31 experimental results summary.Plate AZ31Pin H13Temp. Velocity Pressure ratioµ[°C] [mm/s] [% σ 0 ] mean σ200217.5 % 0.205 0.08615 % 0.179 0.055427.5 % 0.216 0.09015 % 0.167 0.098250217.5 % 0.314 0.09515 % 0.320 0.044427.5 % 0.265 0.06915 % 0.213 0.031300217.5 % 0.630 0.05115 % 0.274 0.013427.5 % 0.449 0.01815 % 0.366 0.0540.140.120.100.087.5 % 15.0 % 27.0 %[mm/s] v21v42Fig. 4. FeP04 parameters interaction effects.Fig. 3 compares the Fisher reference values with therelative F 0 value estimated for single and multipleparameter interactions. The comparison shows thatpressure strongly influences friction. In particularFig. 4 shows the parameters interaction effectswhere it is possible to see how an increase of0.08In particular, Fig. 6 compares the estimated Fishervalues for the AZ31 results. It is possible to observethat temperature and contact pressure affect frictionin the tested ranges, while velocity does not. Inparticular, the most affecting parameter istemperature which causes a friction raise, while anincreasing contact pressure causes a frictionreduction as shown in Fig. 7.An explanation of the temperature influence can begiven as follows: as the temperature increases, thematerial surface (at peaks roughness level) is moredeformed. This means that, for the same pressureratio, the real contact area and, consequently, the

force needed to move the pin (friction force)increase. This is in agreement with the experiencewhere hot forming causes higher friction forces.Observing the contact surfaces after the tests, it wasfound a deposition of AZ31 on H13. In this situationthe material slides on itself so varying the frictionconditions. This fact can explain the friction growthhighlighted in B interval of Fig. 5. This implies theneed of lubricant in AZ31 warm forming to keep theintegrity of the die and workpiece surfaces.Pareto Chart of the Standardized Effects(α=0.05)common trend between the two curves with a moreevident friction minimum at the temperature of300°C and at a pressure of 15% of the yield stress.4.3 Considerations on data scatteringComparing the standard deviations reported inTable 2 and Table 3, it is possible to observe that thetests conduced on AZ31 sheets are affected by anhigher scattering. This scattering decreases as thetemperature increases and should be correlated witha problem in temperature control system.TpT*pT*v*pvp*vT*vFo limitFoStandardized EffectFig. 6. Pareto Chart for AZ31 results.5 CONCLUSIONS AND FUTURE STUDIESThe present study allowed to identify the frictionmost affecting parameters while forming AZ31 andFeP04 sheets. Furthermore, it was identified howthese parameters influence friction.Further studies will be conducted to understand thedependence of friction on pressure, temperature andlubrication especially for AZ31 sheets.0.500.350.200.500.350.20µ[°C] T200T 250300200°C 250°C 300 °C0,550,500,450,400,350,300,250,20Interaction Plot (data means) for µ7,5 % 15,0 %0.50[% σ ] p07.5p15.00.350.20[mm/s] vv214221 mm/s 42 mm/sFig. 7. AZ31 parameters interaction effects.[°C]T2003007.5 % 11.5 % 15.0 % 18.5 %p [% σ 0]Fig. 8. Results of the tests at 200°C and 300°C and slidingspeed equal to 21 mm/s.To better understand the contact pressure influence,additional tests were performed. Namely 11.5% and18.5% of the yield stress at the temperature of 200°Cand 300°C and a sliding velocity of 21 mm/s wereconsidered. The result comparisons (Fig. 8) show aREFERENCES1. H. Sofuoglu, J. Rasty, On the measurement of frictioncoefficient utilizing the ring compression test, TribologyInternational, 32 (1999) 327-335.2. R. Combarieu, , P. Montmitonnet, Effect of additives onfriction during plane strain compression of aluminiumstip test, Wear, 257 (2004) 1071-1080.3. E. Yoon, H. Kong, O. Kwon, J. Oh, Evaluation offrictional characteristics for a pin-on-disk apparatus withdifferent dynamic parameters, Wear, 203-204 (1997)341-349.4. H. Cho, T. Altan, Determination of flow stress andinterface friction at elevated temperatures by inverseanalysis technique, J. of Mat. Proc. Tech., 170 (2005)64-70.5. Y.C. Lam, S. Khoddam, P.F. Thomson, Inversecomputational method for constitutive parametersobtained from torsion, plane-strain and axisymmetriccompression test, J. of Mat. Proc. Tech., 83 (1998) 62-71.6. F. Klocke, G. Messner, C. Giardini, E. Ceretti, FEsimulationof micro-tribological contacts in coldforming: experimental validation, In: 2° Int. Conf. onTribology in Manuf. Proc, ICTMP, Nyborg – Denmark(2004), 395-402.7. E. Ceretti, A. Fiorentino, A. Attanasio, C. Giardini,Influence of die material and roughness on frictioncoefficient in cold forming, In: Proceedings of the 8 thAITeM conference, CET, Montecatini Terme (PT) – Italy(2007) 115-116.8. E. Ceretti, C. Contri, C. Giardini, Study on microtribologicalcontacts in cold forming: simulations andexperimental validation, In: Proceeding of 7th A.I.Te.MConference, CET, Lecce – Italy (2005)117-118.9. F.K. Chen, T.B. Huang, Formability of stampingmagnesium-alloy AZ31 sheets, J. of Mater. Proc. Tech.,142 (2003) 643-647.

Effects of lubricant and lubrication parameters on friction during hotsteel forgingE. Daouben 1,2 , A. Dubois 1 , M. Dubar 1 , L. Dubar 1 , R. Deltombe 1 ,N.G Truong Dinh 2 , L. Lazzarotto 31 Laboratoire d’Automatique, de Mécanique et d’informatique Industrielles et Humaines, UMR 8530,University of Valenciennes, F-59313 Valenciennes Cedex 9, FranceURL: www.univ-valenciennes.fr/LAMIH e-mail: andre.dubois@univ-valenciennes.fr2 CONDAT Lubrifiants –Avenue Frédéric Mistral, F-38670 Chasses sur Rhônes, France3 CETIM – Etablissement de Saint Etienne – 7 rue de la presse – BP 802, F-42952 Saint Etienne Cedex 9,FranceABSTRACT: The aim of the present work is to improved comprehension of wear phenomenon during hotsteel forging with sprayed lubricants. The study reported in this paper uses a specific friction test, called theWarm Hot Upsetting Sliding Test (WHUST), which reproduces hot forging contact conditions. Wear markers–such as friction force/normal load ratio or sliding distance before the first scratch– are proposed tocharacterize the contactor wear at microscopic and macroscopic scales. Friction tests are performed on1100°C heated specimens to characterize the influence of graphite based lubricant film thickness and particlesizes on friction and wear in the flash zone of a nitrided steel die.Key words: Hot forging; steel; friction test, wear, lubricant, graphite1 INTRODUCTIONThe choice of an efficient lubricant is still a criticalpoint in the development of a metal forming process.A bad lubricant, or a good lubricant appliedaccording to bad conditions, may lead toprematurely tool wear, and consequently have adramatic influence on production. The aim of thepresent work is to improve the comprehension ofwear phenomenon during hot steel forging withsprayed lubricants [1]. The study uses a specifictribological testing device: the Warm Hot UpsettingSliding Test (WHUST). The WHUST simulates hotforging contact conditions between specimensheated up to 1100°C and contactor heated up to200°C. Friction tests are performed so that WHUSTcontact pressure, sliding velocity, contactor slidingvelocity, contactor and specimen temperatures aresimilar to industrial ones. Before performing thefriction tests, lubricant is sprayed on contactorsurface. Then the contactor slides against thespecimen with a constant penetration, leaving aresidual deformed track on its surface. MainWHUST results are tangential and normal loads oncontactor, surface roughness and chemicalcompositions on both specimen and contactorsurfaces. A methodology is proposed to analysethese results and provide “wear markers” whichcharacterized the contactor wear at microscopic andmacroscopic scales. The methodology is the appliedto the hot forging of a steel tripod using water basedgraphite lubricants. The effect on friction oflubricant film thickness, and solid lubricant particlesize are analysed.2 WHUST PRINCIPLEDuring the test, the contactor penetrates thespecimen and moves along specimen surface with aconstant penetration and sliding velocity. TheWHUST parameters are the specimen and contactortemperatures, the contactor velocity, the contactorgeometry and the contactor penetration within thespecimen (figure 1). Those test parameters areadjusted so that the test temperatures, the contactpressure at contactor/specimen interface, thespecimen plastic strain and the sliding velocity areclosed to those of the studied process.Contactors are machined from actual industrial tools

and specimens are parts of actual industrial workpieces. That specificity enables the WHUST tosimulate physical and chemical contact conditionsencountered in the studied forming process: thelubricant, surface roughness, chemical propertiesinvolved in the test are the same as those of theprocess [2].3 WHUST RESULTSFig. 1. Overview of the W.H.U.S.TWHUST direct results are normal and tangentialforces on contactor, specimen and contactor surfaceprofiles. From these results we can definemacroscopic or microscopic “wear markers”.3.1 Macroscopic markersThree parameters are defined: the lubrication indexLi, the contactor roughness Ra and the critical lengthLc [3].3.1.a Lubrication index LiThe lubrication index corresponds to the ratio of thetangential and normal forces: Li = F t /F n . This indexprovides quick and reliable information on theevolution of friction. Correlation betweenlubrication index Li and the Coulomb’s coefficientof friction can be found in [4].3.1.b Contactor roughness RaThe mean surface roughness Ra is measured oncontactor contact surfaces before and after each test.It provides information on the occurrences ofadhesive and/or abrasive wear.3.1.c Critical length LcPlastic strains generate a residual track on thespecimen surface during the tests. Scratches can beobserved on this track when the lubricant filmbreaks down. The critical length Lc corresponds tothe dimensionless distance measured along the track,from the beginning of the contact to the point wherethe first scratch is observed. Lc equals zero whenscratches appears as soon as the contactor comes incontact with the specimen; Lc equals 1 when noscratches are observed. This marker characterizesthe ability of lubricant to delay the first metal-tometalcontact between tool and work piece.3.2 Microscopic markerMicroscopic markers are observed on contactorcontact surfaces. They derive from SEM-EDSanalyses.3.2.a SEM analysisSEM pictures are taken on contactor surface at amagnification of 200. These pictures show thepresence or absence of material deposits oncontactor surface (presence of adhesive wear area,oxide scale…). It allows quantifying the size ofthese deposits.3.2.b EDS analysisThe SEM observations are completed by EDSanalyses in order to determine the chemicalcomposition of the material deposits. First, amapping is taken before the friction test in order toexamine lubricant structure. Then, two mappings arerealised after the test, one to determine the presenceand nature of oxide scale [5] and the other toquantify the residual lubricant on contactor surface.4 APPLICATION4.1 Forming process: hot forging of a tripodThe hot forging of a tripod (work piece with threesymmetry planes at 120°) is studied in the presentwork. The process involves two forming sequences:a preform followed by the finishing. Both are carriedout using a mechanical press with a maximumvelocity of 1m/s [6]. The present study focuses onthe lubricant behaviour in the flash zone, at the endof the preform sequence.Dies are made of nitrided tool steel and the tripod isin medium alloyed steel. Dies and work piecetemperatures respectively equal 200°C and 1100°Cwhen they come in contact. A graphite suspension inwater lubricant is sprayed on the dies before formingto reduce friction and limit wear [7]. Measurementsoperated on industrial dies show that the graphitefilm thickness varied from 10 to 40 µm, dependingon the angle between the lubricant spray nozzle andthe surface of the tool. A finite element simulationof the process shows the contact pressures equalled

to 190 MPa and sliding velocities equalled to 60mm/s in the flash zone [2]. Each testing condition isperformed three times, with three “new” contactors.4.2 Effect of lubricant film thicknessFour film thicknesses are simulated on the WHUST:• 0 µm: dry lubrication,• 10 µm and 40 µm, which correspond to theminimum and maximum thicknessesmeasured on the industrial tools,• 30 µm which is the mean film thicknessmeasured on the flash zone of the dies.Table one sums up the main results for themacroscopic wear markers. Contactor surfacesbefore and after the friction tests are presented onfigures 2 and 3.Bonding elements used in the formulation of thelubricant have a great influence on the structure ofthe graphite layer on tool surface. In the presentstudy, for thin lubricant thickness, the graphite layeris almost only composed of powder, withoutparticular structure (figure 2b). The layer structureevolves to a “honeycomb” structure when the filmthickness equal 30µm (figure 2c), and to a stratifiedlayer for thicker films (figure 2d).WHUST performed without lubricant lead to strongmaterial deposit on contactor surface. Thelubrication index is then very high and scratchesappears on the work piece surface after a shortsliding distance (Lc = 0.5). SEM-EDS analysisshows that the material deposit is oxide scalecoming from specimen surface.Large oxide scales particles are observed oncontactor surface when performing test with a 10µmlubricant film thickness (figure 3b). The lubricationindex Li is 28% lower than in dry contact condition,and the critical length Lc increase from 0.5 to 0.87.The graphite layer reduces friction and delay thefirst direct metal-to-metal contact, but is not bondedsufficiently to the surface to avoid scratchoccurrence.Friction tests operated on thicker lubricant filmspresent less material deposit on contactor surface(figures 3c-d). SEM-EDS show that the surface ismostly cover with residual graphite. No scratch isobserved on the specimen surfaces and thelubrication index decreases with the increase of filmthickness.As a conclusion, the film thickness in the flash zonemust by greater than or equal to 30µm to limit toolwear and reduce friction. Nonetheless, it should benotice that the aim of the flash being to slow downmaterial flow in order to guaranty a good die filling,a strong reduction of friction is not always wanted.So, for the studied process, a film thickness equal to30µm is the better compromise between surfacedefect protection and friction control.Fig. 2. SEM pictures of contactor surface before friction testa) dry friction, b) 10 µm, c) 30 µm, d) 40 µmFig. 3. SEM pictures of contactor surface after friction testsa) dry friction, b) 10 µm, c) 30 µm, d) 40 µmTable 1. Evolution of Li and Lc macroscopic markers infunction of lubricant film thicknessLubricant filmthickness0 µm 10 µm 30 µm 40 µmLubrication index Li 0.81 ±0.07 0.58 ±0.04 0.39 ±0.03 0.26 ±0.02Critical length Lc 0.50 ±0.04 0.87 ±0.04 1.00 ±0.00 1.00 ±0.00

4.3 Effect of lubricant particle sizeCommon particle sizes in graphite lubricants rangefrom 2 to 50 µm [8]. To study the lubricant gradingeffect, two lubricants are tested, one with mediumparticle sizes (25-30 µm), another with large particlesizes (50µm). Both lubricants have the samecomposition (same bonding agent), and are sprayedon the contactor so that their thickness equals 30µm.Table 2 exhibits the change of tribological behaviourwhen increasing the particle sizes of the lubricant:the lubrication index increases from 0.39 to 0.48 andthe critical length becomes lower than 1.Figure 4 presents the surface of the contactor justbefore the test. Blue areas are for carbon and red foriron. Lubricant with medium particle sizes has amore homogeneous covering on tool surface. Zoneswhere almost no graphite is present are observed oncontactor surfaces sprayed with the large particlesize lubricant. As a consequence, the wholecontactor surface is not protected, direct metal-tometalcontact occurs between specimens andcontactors and scratches appear on specimensurface. Scratches lead to an increase of the frictionforce.As a conclusion of these tests, when large particlesizes are used, the lubricant film layer has to bethicker so that the bonding agent can react to form ahomogeneous graphite layer.5 CONCLUSIONSAn original methodology focused on the analysis ofwear phenomena in hot forging of steel has beenpresented. A prototype testing device calledWHUST reproduces industrial hot forging contactconditions. WHUST results are analysed usingmicroscopic and macroscopic wear markers. Testsoperated with different lubricant film thicknessesand different lubricant particle sizes have shown thatthe efficiency of the lubricant film is linked tostructure of the graphite layer more than thelubricant thickness.These results are very promising and will help tounderstand wear mechanisms in a very near future.The next steps of this study will be theunderstanding of the behaviour of graphite freelubricant for hot forging and the testing of industrialdies at the end their lifecycle.Table 2. Evolution of Li and Lc macroscopic markers infunction of lubricant film thicknessLubricant particle size 25-30 µm 50 µmLubrication index Li 0.39 ±0.03 0.48 ±0.02Critical length Lc 1.00 ±0.00 0.90 ±0.04Fig. 4. SEM-EDS pictures of contactor surface before frictiontests a) EDS, medium grading, b) EDS, large gradingACKNOWLEDGEMENTSThe present research work has been supported by the EuropeanCommunity, the Délégation Régionale à la Recherche et à laTechnologie, the Ministère de l’Education Nationale, de laRecherche et de la technologie, The Région Nord-Pas deCalais, the Centre National de la Recherche Scientifique,CONDAT SA and CETIM, ASCOFORGE and Forges deCourcelles. The authors gratefully acknowledge the support ofthese institutions and industrial partners.REFERENCES1. O. Barrau, C. Boher, R.Gras, F. Rezaï-Aria, ‘Analysis ofthe friction and wear behaviour of hot forging’, Wear 255(2003) 1444-1454.2. R. Deltombe, N. Morgado, M.Dubar, A.Dubois and L.Dubar, Hot forming of aluminium: a new methodology toidentify friction parameters, 7 th ESAFORM, Norway –Trondheim (2004) 253-256.3. E. Daouben, L. Dubar, M. Dubar, R. Deltombe, A.Dubois, N. Truong-Dinh, L. Lazzarotto, ‘Friction andwear in hot forging of steels’, Proceedings of the Int.Conf. ESAFORM 2007, Zaragoza, Spain (2007)4. L. Lazzarotto, L. Dubar, A. Dubois, P. Ravassard and J.Oudin, ‘Identification of Coulomb’s friction coefficientin real contact conditions applied to a wire drawingprocess’, Wear, 211 (1997) 54-635. B. Picqué, P.O. Bouchard, P. Montmitonnet and M.Picard, ‘Mechanichal of iron oxide scale: Experimentaland numerical study’, Wear, 260 (2005) 231-242.6. S.Sheljaskow, Tool lubricating systems in warm forging.J. Mater. Process. Technol., 113 (2001) 16-21.7. T.Iwama and Y. Morimoto, Die life and lubrication inwarm forging. Journal of Materials ProcessingTechnology 71 (1997) 43-48.8. R. F. Deacon and J.F. Goodman, ‘Spreading Behavior ofwater based graphite Lubricants on Hot Die Surfaces’,CIRP 55/1 (2006) 299-302.

Behaviour of oxide scales in hot steel strip rollingC. Grenier 1,2 , P.-O. Bouchard 1 , P. Montmitonnet 1,* , M. Picard 21 Ecole des Mines de Paris - ParisTech, UMR CNRS 7635, BP 207, 06904 Sophia-Antipolis Cedex, FranceURL: www.cemef.ensmp.fr e-mail: Claire.Grenier@ensmp.fr ;Pierre-Olivier.Bouchard@ensmp.fr ;Pierre.Montmitonnet@ensmp.fr2ArcelorMittal, R&D Centre, BP 30320, 57214 Maizières-les-Metz Cedex, Francee-mail: Michel.Picard@arcelormittal.comABSTRACT: The behaviour of oxide scales in the finishing Hot Strip Mill is simulated by the hot PlaneStrain Compression Test (PSCT). Compared with the ideal case of homogeneous plastic co-deformation ofthe oxide layer and the underlying metal, different types of defects are described: delamination at the interfaceor within the oxide layer; interfacial plastic instability due to the jump of the mechanical properties;perpendicular, through-thickness cracks where the axial strain parallel to the interface dominates, followed bymicro-extrusion of metal between the fragments; oblique cracks followed by sliding along the lips, whereshear dominates. The Finite Element Method (FEM) is used to bring elements of interpretation, as to whichconditions determine each mechanism. Conclusions for the behaviour in hot rolling are sketched.Key words: Oxide scales, hot rolling, Plane Strain Compression Test, PSCT, Finite Element Modelling1 INTRODUCTIONIn hot rolling of steels, the rolled material in factconsists of thin layers of a ceramic, the complex ironoxide layers (20 – 50 µm thick), on hot metal. Thedifference in hardness and ductility of these twomaterials [1] often leads to oxide cracks of variouskinds, through-thickness or interfacial (oxidespalling) [2-5]. The present paper focuses onthrough-thickness cracks.In a previous work [6-7], fracture occurring justbefore bite entry had been studied bothexperimentally (hot bending test) and theoretically(FEM). Cracks open wherever superficial tensilestresses occur, then they may open wide in the bitedue to strip elongation. If the pressure is highenough, "micro-extrusion" of fresh hot metal takesplace through the open cracks. The interfacebecomes wavy as in Figure 1 ("rolled-in scale"), adefect which may become visible after pickling andremain even after cold rolling. A numericalparametric study resulted in a damage risk chart,where thicker oxides (e.g. due to high temperatures)were found to be most detrimental, unless sufficientplasticity allows them to keep up with part of theelongation of the strip. In terms of material data, theoxide-to-steel hardness ratio and the temperaturedependentand strain-rate-dependent oxide fracturestress were found most important.Figure 1: oxide cracks and wavy oxide – metal interface.Above: top view [2] shows an array of cracks normal to rollingdirection (RD). Below: cross section (this work).The present work aims at extending these notions tocracks opening in the roll-strip contact (bite). Thereare few data available on the morphology, natureand origins of this particular category.yzxxzyx

2 EXPERIMENTAL2.1 PSCT set-up and procedurezyxFigure 2: PSCT test rig (left) and procedure (right).PSCT consists in upsetting a strip between two flatdies (figure 2 left). Here, the oxidation of steel stripsamples is made in situ, by allowing temporarily anoxidative atmosphere in the protective glass vessel(figure 2 right). The oxidation temperature is 900°Cin all tests, i.e. close to entry temperature in thefinishing Hot Strip Mill (HSM). The oxidation timeis varied to give oxide thickness between 10 µm and100 µm. The system is then brought up or down tothe mechanical test temperature. After temperatureequilibration, the test is performed in a fraction of asecond; in the tests reported, no lubricant was used.Then the sample is allowed to cool freely in N 2 .2.2 MaterialsThe steel strip is an ultra-low carbon, DWI deepdrawingsteel, with 0.015%C, 2.29% Mn, 0.23% Si(atomic concentrations). The coupons are 5 mmthick, 50 mm wide and 62 mm long.The die material is yttria-toughened zirconia; diesare 70 mm long and 12 mm wide. Rough dies (Ra =3.6 µm, isotropic) are compared with smooth dies(Ra = 0.42 µm, isotropic); on occasion, grooved diesare used to simulate damaged (“banded”) rolls; thegrooves are in the punch width direction x,equivalent to the rolling direction (RD).2.3 Experimental plantemperature (°C)11001000900800700600500400N 2 N 2Oxidation(N 2 +O 2 +H 2 O)Deformation0 500 1000 1500time (s)Oxide thickness: 10µm, 25 µm, 50 µm, 75 µm.PSCT Temperature: 800°C, 900°C and 1050°CStrain: ε = 0.2 ; 0.4 ; 0.6 ; 0.8.Strain rate: 0.1 s -1 , 1 s -1 and 10 s -1 .3 EXPERIMENTAL RESULTS3.1 General observations: cracks and interfaceabFigure 3: Top view: vertical, normal cracks on the side of aPSCT indentation. (a): enlarged view of the framed area in (b);the white bar is 1 mm.1ε , & ε = 1.s− , T = 900°C, 50 µm oxide.Smooth die, = 0. 4yFigure 4: SEM picture of a cross-section in the flow direction(x = punch width direction ↔ RD). Same conditions. Metal iswhite, oxide is light grey.Figure 3 shows a typical aspect in top view, an arrayof cracks perpendicular to the major flow directionx, very similar to cracks opening before the bite [6-7]. Figure 4 shows they run through the oxidethickness, with a uniform density. Their origin hereis not the oxide bending ahead of the bite as in [6-7]but probably the flow of the underlying metalputting the oxide in tension. It might also be due topunching by die roughness peaks; this will bediscussed using numerical simulation in paragraph 4.Spalling (figure 4) may be due to sample polishingbefore microscopic observation; but slight interfacewaviness suggests spalling or crushing during thetests, followed by micro-extrusion, as in [6-7].This observation answers one of the questionsbehind this work: crack array formation may go onin the contact - PSCT is known to be a goodsimulator of strip rolling. The subsequent evolutionseems very similar to pre-bite cracks.In the case of pre-bite cracks, the density could berelated to oxide thickness, oxide fracture stress andinterface shear stress [6-8]. We expect numericalsimulation to help determine the parametricdependencies for the present, in-bite opening cracks.It has been found occasionally that cracks may notalways be normal to the surfaces. Figure 5 shows acase where oblique cracks, followed by rotation offragments, have occurred near the edge of a die(note that rotation may have been facilitated by thepresence of a thick lubricant film in this particularx

case). The same pattern has been found by [9] on hotrolled strips. This proves the relevance of thisphenomenon, tentatively attributed to the presenceof significant shear stress (die edge / friction) whichinduces the rotation of principal axes.zx3.3 Interface3.3.a Delamination / spallingThis is another failure mechanism whereby oxidefragments may be spalled off the sample surface; inrolling, such fragments may then be embeddedinside the strip surface by the contact pressure.zyFigure 5: oblique cracks (top) near the PSCT die edge,(bottom) on a rolled strip [9].3.2 Roughness transferFigure 6: comparison of the oxide surface state after PSCT1with rough / smooth dies. Die width 12 mm. ε = 0. 4 , & ε = 1.s− ,T = 900°C, oxide thickness 50 µm.Figure 8: two examples of delamination. Left: interfacialdelamination on the flank of a groove. Right: delaminationwithin the oxide layer.In figure 8 (left), bending of the sample in the flankof a grooved die (representing the “roll banding”defect, i.e. peeling of an orthoradial strip of rolloxide) has resulted in a normal, through-thicknesscrack which has bifurcated along the interface; insuch situations, the formation and embedding of afragment becomes highly probable. The die grooveis oriented in the die width direction, equivalent tothe rolling direction. Such a crack would thus belongitudinal, contrary to those shown above. Figure8 (right) shows delamination within the oxide. Linesof pores have occasionally been found in oxidelayers, parallel to the interface; they may be theorigin of such defects. Yet delamination might alsohave taken place at the interface, with subsequent reoxidationduring cooling in imperfectly pure N 2 .Here again, fragmentation and embedding of thespalled layer in the next rolling stand is inevitable.Figure 7: cross-section of the samples shown in figure 5,showing the smooth metal (white) – oxide (grey) interface,whatever the die roughness.Another possible defect is interface waviness due totool roughness printing, in particular when rolls areseverely worn. Figure 6 shows that the oxidesurface, to a large extent, takes the roll roughness,which suggests a certain degree of plasticity of theoxide, but the oxide – metal interface remainssmooth (figure 7). This may not be a generalconclusion however, certainly dependent (i) on theratio of the tool roughness to the oxide thickness,and (ii) of the temperature-dependent mechanicalproperties of the oxide (toughness, oxide-to-metalyield stress ratio). Here again, numerical simulationcan contribute in the study of these parameters.3.3.b Interface instabilityIn a series of tests devoted to varying strain rate, asinusoidal interface waviness has been observed atthe lowest strain-rate, decreasing and disappearingas ε & increases (figure 9).Figure 9: Interface waviness without cracking. Strain rateincreases from top to bottom (0.1, 1 and 10 s -1 ).There is no evidence of any crack / micro-extrusionphenomenon, hence the interpretation by plastic codeformationinstability (see [10] e.g.). It is not surethat this can occur in strip rolling, since it has been

found here only in a low strain rate range; yet itmight occur at higher speeds under differentconditions (temperature…).4 THEORETICAL INTERPRETATION4.1 Model descriptionThe Forge2005® FEM software has been used forthis study. Its description can be found in [6,7],together with the added through-thickness crackopeningalgorithm (based on a critical fracturestress). Only such cracks are addressed here,although a delamination capability has also beenadded (Cohesive Zone Modelling approach).4.2 An application to interpretation of crack originFigure 10 shows a crack opening simulation in thePSCT at 1000°C and 1 s -1 , with a critical fracturestress of 200 MPa for the oxide. The oxide and themetal yield stresses are respectively given by:σ ( & ε0.00299. T ( K ) 0.2230.159MPa ) = 69 . exp . ε . ( 3. ) (1)⎛ 3340 ⎞σ ( &⎜ε( )⎟⎝ T K ⎠MPa 0.22 0.09) = 8.5 . exp⎜⎟ . ε . ( 3. ) (2)The die-oxide friction factor is m = 0. 08 . The oxidemetalinterface is assumed perfectly adherent.Figure 10: PSCT FEM modelling and crack formation.Under these conditions, cracks form either under dieedges (figure 10, left) due to the stress singularity, orat asperity tops (figure 10, insert right), but never onflat parts of the die. Looking in more details, withthe roughness peak height chosen (3 µm, to becompared with oxide thickness 50 µm), fracturedoes not occur because of the stress singularity atpeak apex, since full penetration of the asperity intothe plastic oxide occurs before crack opens. Thecritical tension is in fact reached later on, when theflow of the underlying metal shears the interface andputs the oxide in tension. In another simulation witha periodic roughness covering the whole die width(Ra = 0.3 µm), cracks appear periodically, but thewavelength is much larger than the roughnesswavelength. Study of the influence of parametersand comparison with experiments are in progress.The effect of friction and shear stress on obliquecrack formation (as in figure 5) is also under study.5 CONCLUSIONSExperimental results present a number ofdeformation and fracture phenomena which occur inoxidized metal / tool contact; their relevance towardsinterface defects in hot strip rolling has beencommented. A first example of application of multibodyFEM to the analysis of the origins of theseoxide and interface defects has been presented.ACKNOWLEDGEMENTThe authors thank ArcelorMittal (Arcelor Research S.A.) forfinancial support and authorization to publish the results.REFERENCES1. G. Vagnard and J. Manenc, Etude de la plasticité duprotoxyde de fer et de l’oxyde cuivreux. Mem. Et. Sci.Rev. Met. LXI 11 (1964) 768-776 (in French).2. Y.H. Li and C.M. Sellars, ‘Modelling deformation ofoxide scales and their effects on interfacial heat transferand friction during hot steel rolling’, Proceedings of the2 nd Conf. Modelling of Metal rolling processes, London(1996) 192-2013. M. Krzyzanowski and J.H. Beynon, 'Oxide behaviour inhot rolling', in: Metal forming science and practice, ed,J.G. Lenard, Elsevier, Amsterdam (2002) 259-2954. M.F. Frolish, M. Krzyzanowski, W.M. Rainforth andJ.H. Beynon, 'Oxide scale behaviour on aluminium andsteel under hot working conditions', J. Mater. Process.Technol. 177, 1-3 (2006) 36-405. M. Schütze, 'Mechanical properties of oxide scales',Oxid. Met. 44, 1-2 (1995) 29-606. B. Picqué, Experimental study and numerical simulationof oxide scales mechanical behaviour in hot rolling, PhDThesis, Ecole des Mines de Paris (2004)7. B. Picqué, P.-O. Bouchard, P. Montmitonnet and M.Picard, 'Mechanical behaviour of iron oxide scale:experimental and numerical study', Wear 260 (2006) 231-2428. D.C. Agrawal and R. Raj, 'Measurement of the ultimateshear strength of a metal-ceramic interface', Acta Met.37, 4 (1989) 1265-12709. F. Platteau, G. Lannoo and D. Espinosa, 'Control of stripsurface quality during hot rolling', Internal Report, CRM(2007) (personal communication).10. S.L. Semiatin and H.R. Piehler, 'Formability of sandwichsheet materials in Plane Strain Compression and rolling',Met. Trans. A 10A (1979) 97-107

Modelling of friction with respect to size effectsZ. Hu 1 , F. Vollertsen 11 BIAS Bremer Institut fuer Angewandte Strahltechnik – Klagenfurter Str. 2, D-28359 Bremen, GermanyURL: www.bias.dee-mail: hu@bias.de; vollertsen@bias.deABSTRACT: In this paper a size dependent FEM-simulation for sheet metal forming was realized applyingthe size dependent friction functions acquired from strip drawing tests [1]. The software ABAQUS 6.6.3 wasused to simulate the strip drawing tests and deep drawing process in different process dimensions. A 2-dimensional model was built for strip drawing test and a 3-dimensional model was built for deep drawingprocess. Both processes were experimentally carried out in different dimensions. A constant frictioncoefficient as well as a friction function, which shows a dependence of friction coefficient on the contactpressure, was applied in the simulation. The simulated punch force vs. punch travel curves were thencompared with the experimental curves. A discussion about the difference between these curves withconsideration of tribological size effects will be given.Key words: Scaling, Tribological size effects, Friction1 INTRODUCTIONWhen downscaling the size of the work piece tomicro forming, not all parameters can be changedaccording to the rule of similarity, e.g. grain size.This causes the so called size effects, i.e. theoccurrence of unexpected results concerning theforming force of the forming limit [1]. Since deepdrawing is essentially affected by the frictionbetween the work piece and tools [2, 3], which isalso affected by the size effects [4, 5], thetribological size effects in sheet metal forming wereinvestigated in our former work [6, 7] and the sizedependentfriction functions were acquired. The aimof this work is to realize size dependent FEMsimulationfor deep drawing applying the acquiredfriction functions.2 TRIBOLOGICAL SIZE EFFECTS2.1 Scaled experimentsCompared to usual deep drawing there is notangential force F t at flange area in strip drawing, seefigure 1.Fig. 1. a) Strip drawing testb) Deep drawingThis difference makes it easier to find the relationbetween the punch force and the friction coefficient.Thus the friction function method was developedonly for the strip drawing, which can be used later toidentify the tribological size effects in deep drawing.The effective friction function method yields afriction coefficient from deep drawing (accordingfigure 1b), having some disadvantages concerningthe dependence on the contact pressure. Bothprocesses were carried out with 5 different punchdiameters in this investigation. They are 50, 20, 10,5 and 1 mm. According to the theory of similarity,all process parameters are kept constant whilealmost all geometrical parameters of tools and work

pieces are scaled by the same scaling factor. Forexample the thickness of the work piece has thesame ratio to punch diameter in each experiment [8].Al99.5 is used as work piece material in everyprocess dimension. The properties of the materialare also affected by the size effects [9, 10]. For thedetermination of the friction function and the frictioncoefficient in this work the flow stress is required.Thus, the flow curves of Al99.5 in each thicknesswere acquired through tensile tests and show clearlydifference from each other, see figure 2. In order toavoid the influence of the surface quality on thefriction, all tools as well as blanks used in this workhave nearly the same roughness respectively.mm is described, which illustrates a tribological sizeeffect within this investigation.The friction functions can be expressedmathematically through an exponential form:μ = C1 + C2⋅ exp( −P⋅C4) + C3⋅ exp( −P⋅C5) (1)Where µ = friction coefficient, P = contact pressure,C 1 -C 5 = coefficients (They are listed in table 1).Table 1. Coefficients of friction functionsPunch diameter C1 C2 C3 C4 C5[mm]50 0.000 0.159 0.130 0.871 0.00720 0.000 0.180 0.130 0.571 0.01110 0.125 12.00 0.449 7.759 1.3465 0.116 0.319 0.032 3.571 2.3481 0.000 0.188 0.180 0.687 0.0102.3 Effective friction coefficientAccording to theory of Storoschew [11] the frictioncoefficient in deep drawing can be evaluated fromthe measured maximum punch force for eachprocess dimension [12], see figure 4.Fig. 2. Flow stress of Al99.5 in different thicknesses2.2 Friction functionThe punch force was measured in strip drawing testsand was used later in the calculation modeldescribed in [6]. For each punch diameter, a frictionfunction is determined. A graphic display of thesefriction functions is shown in figure 3.Fig. 4. Effective friction coefficients from deep drawingIn this theory there is only one friction coefficient.Thus the friction coefficient calculated using thismethod is call effective friction coefficient. By theseeffective friction coefficients the tribological sizeeffects can also be shown: with decreasing theprocess dimension the friction increases.3 FEM-SIMULATIONFig. 3. Friction functions from strip drawing testsA difference with a factor of about 1.5 between thefriction functions with punch diameter of 1 and 503.1 Scaled strip drawing testsIn this work the software ABAQUS 6.6.3 was usedto simulate both forming processes. A 2-dimensional

model was created for the strip drawing test, inwhich the tools were defined as analytical rigid lineand the blank was defined as deformable object. The4-node bilinear plane stress element CPS4R wasused to mesh the blank. Within the thickness of theblank there are four elements. For simulation of theexperiments with blanks in different thicknesses, theflow stress curves of Al99.5 in different thicknesseswere applied respectively in the simulation model.Using the normalization described in [12] thesimulated punch force vs. travel curves for stripdrawing with punch diameters of 1 and 50 mmshown in figure 5 were obtained.reason for this might be the assumption in thecalculation model: The normal pressure at the radiusis uniform [1]. However, the simulation shows twolocal contact zones, see figure 6. The pressure onthese two zones is much higher than the uniformdistributed pressure assumed in the calculationmodel. Since the friction coefficient depends on thenormal contact pressure, if the pressure on thecontact surface is not right, the incorrect frictioncoefficient will be used. Thus the simulated punchforce differs from the experimental one.Fig. 6. The local contact zones at radius of the die3.2 Scaled deep drawingFig. 5. Comparison of simulated and experimental punch forcevs. punch travel from strip drawing testsFor changes of the process dimension the simulationprogram can not take into account the change of thesize automatically. Thus the result of simulation formicro forming can not differ from that of macroforming using the conventional simulation method.Applying the friction function obtained above intothe simulation, the simulated curves (the curveFEM-1 for the punch diameter of 1 mm, the curveFEM-50 for punch diameter of 50 mm) show thesame trend as the experimental curves (the curveEXP-1 for the punch diameter of 1 mm, the curveEXP-50 for the punch diameter of 50 mm). Thenormalised punch force for punch diameter of 50mm is lower than that for punch diameter of 1 mm.All curves have the maximum point at thenormalised punch travel of about 0.2. A FEMsimulationwith consideration of tribological sizeeffects was realised.The maximum point of the simulated curve FEM-1is about 11% lower than that of the experimentalcurve EXP-1. While the maximum point of the curveFEM-50 is about 8% lower than that of EXP-50. TheSimilar to the strip drawing a 3-dimensional modelwas created for deep drawing. The effective frictioncoefficients from deep drawing as well as thefriction functions from strip drawing using the samepunch diameter were applied in the FEM-simulationfor each dimension. The comparison of thesimulated and experimental punch force vs. punchtravel for deep drawing with punch diameter of 1mm is shown in figure 7.Fig. 7. Comparison of simulated and experimental punch forcevs. punch travel from micro deep drawingThe maximum punch force of the simulated curveusing the friction function is about 5% higher thanthat using the effective friction coefficient. Both ofthem show a good agreement with the experimentalcurve from beginning till the maximum punch force.Then both simulated curves begin to decrease while

the experimental one shows nearly no trend todecrease. This result confirms our statement in [12],that the thickness of lubricant can result in anincrease in punch force, because in micro deepdrawing the sum of blank thickness and lubricantthickness might be more than the drawing clearance.This effect can not be detected in macro deepdrawing and was also not implemented in the FEMsimulation.For punch diameter of 50 mm, the maximum punchforce of the simulated curve using friction functionis about 400 N lower than the experimental curve,while the one using effective friction coefficientshows a difference of about 1000 N, see figure 8.forming processes with consideration oftribological size effects.• The distribution of contact pressure will be takeninto account in our future work.ACKNOWLEDGEMENTSThe work reported in this paper is funded by the DeutscheForschungsgemeinschaft (DFG) within the project “Modellingof tribological size-effects in deep drawing” (DFG project no.Vo 530/6). The authors would like thank the DFG for theirbeneficial support.Moreover the authors would like thank the institute of MetalForming and Casting (UTG) in Munich in Germany for carryout the tensile test for the Al99.5 in thicknesses of 0.02, 0.1,and 0.2 mm.REFERENCESFig. 8. Comparison of simulated and experimental punch forcevs. punch travel from macro deep drawingThis means the theory of Storoschew is not suited tocalculate the friction coefficient precisely frommeasured punch force in different dimensions.Oppositely the friction functions from strip drawingtests show a good scalability. However, inaccordance to the results in strip drawing, thesimulated punch force vs. punch travel curve doesnot agree with the experimental curve perfectly. Thereason for that might be the distribution of contactpressure, since the local contact zones exist also indeep drawing. Thus in our future work thecalculation model will be enhanced in order to takeinto account the distribution of contact pressure.4 CONCLUSIONS• Tribological size effects were observed in bothscaled strip drawing tests and deep drawing.• The friction functions from strip drawing testscan be integrated into FEM-simulation e.g.ABAQUS. This enables to simulate sheet metal1. F. Vollertsen, Z. Hu, Tribological Size Effects in SheetMetal Forming Measured by a Strip Drawing Test,Annals of CIRP 2006, vol. 55/1, 291-294.2. D.D. Olssen, N. Bay, Prediction of Limits of Lubricationin Strip Reduction Testing, Annals of CIRP 2004, 53/1,231-234.3. P. Becker, H.J. Jeon, C. C. Chang, A.N. Bramley, AGeometric Approach to Modelling Friction in MetalForming, Annals of CIRP 2003, 52/1, 209-212.4. F. Vollertsen, Z. Hu, H. Schulze Niehoff, C. Theiler,State of the art in micro forming and investigations intomicro deep drawing, Journal of Materials ProcessingTechnology, 2004, 151, 70-79.5. N. Tiesler, U. Engel, Microforming – Effects ofMiniaturisation, Proceedings of the 8th InternationalConference on Metal Forming, Eds. Pietrzyk, M.,Kusiak, J., ec al., Kraków, 2000, 355-3606. Z. Hu, H. Schulze Niehoff, F. Vollertsen, Determinationof the Friction Coefficients in Deep Drawing, Processscaling, eds.: Vollertsen F., Hollmann, F., BIAS-Verlag,ISBN 3-933762-14-6, Strahltechnik 24, 2003, 27-347. Z. Hu, F. Vollertsen, Scaled Friction Test Integrated inDeep Drawing, ICTMP2004, Denmark, Ed. Niels Bay,ISBN: 87-91035-12-0, 561-5688. Z. Hu, F. Vollertsen, Tribological Size Effects in SheetMetal Forming, ICTMP2007, Yokohama, Japan, Ed.Akira Azushima, ISBN 978-4-9903785-0-9, 163-1689. A. B. Richelson, E. van der Giessen, Size Effects inSheet Drawing, 9th International Conference on SheetMetal 2001, eds. Duflou, J.R., Geiger, M., et al., 263-27010. A. Messner, Kaltmassivumformung MetallischerKleinstteile –Werkstoffverhalten, Wirkflaechen -reibung, Prozessauslegung-, Eds. Geiger, M., Feldmann,K., ISBN 3-87525-100-8, 199811. M.W. Storoschew, E. A. Popow, Grundlagen derUmformtechnik, VEB Verlag Technik Berlin, 196812. Z. Hu, H. Schulze Niehoff, F. Vollertsen, Tribologicalsize effects in deep drawing, ICNFT2007, eds. F.Vollertsen, S. Yuan, ISBN 978-3-933762-22-1, 573-582

Discrete Element method, a tool to investigate contacts in materialformingI. Iordanoff 1 , D. Richard 2 , S. Tcherniaieff 11 ARTS ET METIER PARITECH, LAMEFIP – Esplanade des ARTS et Metiers, Talence, FRANCEURL: www.lamef.bordeaux.ensam.fre-mail: ivan.iordanoff@lamef.bordeaux.ensam.fr; Serge.tcherniaeff@bordeaux.ensam.fr2 INSA de LYON, LAMCOS – 18-20 Rue des Sciences, 69621 Villeurbanne Cedex, FRANCEURL: lamcos.insa-lyon.fre-mail:david.richard@insa-lyon.fr;ABSTRACT: This paper is devoted to the description of a numerical tool that allows the local study ofcontacts in material forming: Discrete Element Method (DEM). Discrete Element Methods allow the study oflocal properties (cohesion, thermal generation, fractures) on process behavior. This tool is used as moleculardynamics but allows the simulation of much representative volumes. Discrete element method is applied as atool to understand/propose/confirm, physical scenario involved in the contact zone. It is shown in this paperhow such numerical simulations can be used as a complementary tool for forming processes study. Examplesare given on thermal study in cutting process, subsurface damages analysis in abrasion process and weldingjoint characterization in Friction Stir welding.Key words: Discrete Element Method, Abrasion process, Friction Stir welding, Tool Machining.1 INTRODUCTIONDiscrete element method has been widely developedfor rheological study of true discrete materials likesand, powder and granular materials [1]. In the pastten years, their fields of applications have beenextended to heterogeneous materials like concrete,biological materials or foams. Their ability tosimulate multi body behavior is used for problemswhere:- A great number of dissociated elements mustbe taken into account,- A great number of default is encounteredThe main recent fields of applications are multifracturesproblems, where detached elements mustbe taken into account (wear in tribology [2],avalanches in geophysic, milling, grinding …).In material forming or cutting, the contact zonebetween the tool and the working piece is often verydifficult to analyze because:- The affected area has little dimension,- High mechanical, rheological, thermalgradient are involved,- Physical phenomena are highly dynamic.To analyze the contact behavior Godet developedthe third body concept [3]. This included adescription of the formation and movement offragmented particles in the interface region. Tostudy the behavior of the third body inside andoutside the contact, Berthier proposed [4] thetribological circuit which allows the study of masstransfers inside the contact. Based on thistribological circuit and coupling Discrete Elementmodels (DEM) to experimental, but simplified, wearstudies, Fillot et al [5] proposed a set of equationsthat allows a qualitative modeling of wear as a massbalance in the contact area. Iordanoff et al. [6]showed how abrasion process can be studied as aparticular and controlled wear process.This paper first presents the Discrete ElementMethods developed in the special case of materialforming. Then, three examples are given to illustratehow this numerical tool can be used to study localproperties in material forming: investigation of SubSurface Damage in abrasion process, welding jointcharacterization in Friction Stir welding and thermalinvestigation of the tool-chip contact during cuttingoperation.

2 NUMERICAL TOOL2.1 General descriptionThe method described is based on smooth particledynamic Method. The material is considered as a setof discrete particles that moves under a force field.The forces are contact forces to simulate multi bodyinteractions and joint forces to simulate a solid madeof discrete elements. Particle movements arecalculated using an explicit algorithm to integratethe dynamic Newton law. A thermal model can becouple with the mechanical one. Both Conductionthrough contact and thermal energy source due tomechanical dissipation are taken into account. Thetwo next paragraphs briefly described themechanical and thermal part. More details can befound in references [2] and [7].2.2 Mechanical PartThe particle interaction laws define the micromechanical properties of the media. They aredivided into contact forces and joint forces. Everyparticle is subjected to contact force. This force actsonly when two particles geometrically interact. Twoadjacent particles belonging to the same solid arelinked by a solid joint that acts under traction andshear state.δ sδ ncContact Forceδ ntLink ForceFig. 1. Contact and Link force calculation.Contact forceContact force is divided into three parts: repulsion,adhesion (both are energy conservative) and energydissipation. Geometrical interaction δ nc allows theforce calculation (Fig.1).• Repulsion is represented by a linear spring,whose stiffness is K. Repulsive force F r is:Fr= K * δnc(1)• Adhesion has been simplified to a constant γ:Fa= γ(2)• The energy dissipation in the contact is due tothe damping force written as:Fd = 2αK.M &ij* δnc(3)where α is the damping coefficient (

een neglected due to the strong density of thedomain and the higher influence of diffusion.3 APPLICATIONS3.1 SSD Study in abrasion process.zyPeriodical boundariesUpper wallDegradableFirst bodyImposed velocityAbrasive particlesLower wallxFig.2 Simulated domain3.1.a Simulated domainThe studied material is silica. According to fig. 2,the upper first body (Silica piece to be surfaced) isconstituted by spheres linked together by elasticsolid joints. This first body is linked to the upperwall. A normal load is applied to the upper wallwhich is free to move along axis z. The lower firstbody (tool) is simply defined as a lower rigid wallmade of adjacent spheres. A constant velocity alongx is applied to the lower wall. The abrasive particlesare plates composed by 4 adjacent spheres linked byan elastic non breakable solid link.The effect of abrasive size and abrasive quantitythrough the contact are studied. Abrasive size is thesame than silica particles (*1) or twice silicaparticles (*2). Table 1 summarizes the studied cases.Nb of Abra. plates abrasive sizeCase ‘DiscreteBig’ 6 *2Case ‘LayerOfLittle’ 39 *1Case ‘DiscreteLittle’ 19 *1Case ‘L.ayerOfbig’ 24 *2Table 1 :Abrasive layer definition.3.1.b ResultsCalculations are carried out till a certain amount ofsilica particle is removed from the silica volume, bythe effect of broken links. The fixed amount is 150particles. Broken links through the volume areplotted in terms of the distance from the surface.Curves fig. 3 show the results. The thickness isdivided by particle mean radius.Nb/unit of Volume10 010 -110 -210 -31 : DiscreteBig2 : DiscreteLittle3 : LayerOfLittle4 : LayerOfBig12 45 0 10 5 15 10 20 15 25 20 3025Dimensionless ThicknessFig.3 :Number of broken joint through the thicknessThe number of broken joints decreases exponentiallywith the depth. It greatly depends on abrasivegeometrical properties and quantity. A little numberof big abrasive particles creates more residual cracksin the material. These results are qualitatively inaccordance with experimental results [9]. This firstsimple study demonstrates how discrete elementsimulations have the ability to simulate theformation of a great number of cracks and the linkbetween sub surface damage and abrasive properties.33.2 FSW joint characterization50+100454035-10V low302520150 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1V high0% 50%100%Fig.4 : quality of mixture in an FSW joint.Simulating friction Stir welding requires thesimulation of material flow, material mixture, toolmaterialcontact and thermo mechanical coupling.D.E.M. could be in that case an interesting tool,because it has been widely used to simulate mixtureof true granular flow. A first simulation is carriedout with a tool that passes through two bodiesinitially in contact (two colours in fig.4). The bodiesare simulated by an assembly of cohesive sphericalparticles. The particles of the two bodies are able tomix thanks to the rotating tool that pass through theinterface. Fig.4 shows the result, in terms of mixturequality for two cases. For one simulation thetranslation velocity has the same order of magnitudethan the rotating velocity. For the other one thetranslation velocity is largely less than rotatingvelocity. It is found that the quality of mixture is

much better when rotating velocity is high comparedto translation velocity3.3 Thermal Tool-Chip Contact investigationPeriodical boundariesAIR/CHIP SURFACEPressureCHIPVelocityCUTTING TOOL SURFACEFig. 5. Tool-chip contact domain in DEMThe chip is modelled as a set of cohesive sphereswith Steel properties. The upper wall represents thecontact between the chip and the ambient air (forconvection process) and the lower wall is the cuttingtool surface (for diffusion process) sliding on thechip. During its generation, the chip faces differentpressure fields from high levels (penetration of thetool in the material) to low levels (evacuation of it).The rheology of the chip greatly varies for each levelof constraint, leading to different velocityaccommodation processes and heat generationlocalisation [7]. The pressure effects on temperatureincrease within the chip and the tool are studied,assuming that all the mechanical dissipative power(from Eq. 3) is converted into heat.Chip thicknessToolLow pressureIntermediate pressureHigh pressure0 10 20 30 40 50 60 70 80Temperature increase °KFig. 6. Temperature profiles in the chip and part of the toolthickness for 3 different applied pressures.Simulations have been carried out for a slidingvelocity of 0.58m/s and three pressures (10, 20 and200Mpa). It shows that the air plays an importantinsulating role: the maximum temperature is at theair-chip interface. The temperature increase of thetool is very small compared to the chip, the diffusionallowing an efficient evacuation of the heat,especially during the high pressure process. Thesefirst results are in accordance with classical thermalstudy of the tool-chip interface.4 CONCLUSIONThe discrete Element model has been presented.Three examples showed the capability of suchsimulations to analyse locally thermo mechanicalproperties of contacts between the tool and thematerial. The studied cases are very simplified butallow finding classical results from the literatures.Generally, to find such results with continuousmodels, both rheological laws and numericalmethods reach high level of complexity. Here, themicro-mechanical model is simple and the behaviouris complex (multi-cracks, high deformation levels,thermo-mechanical interaction). D.E.M is apromising tool for the material forming field, wherematerials are submitted to very high stress, strainand thermal gradients and where rheologicalcharacterisation is still a true challenge. The maindefault of such method is the limitation in terms ofthe number of simulated particles (10 4 to 10 6 ) due tocalculation duration. Coupling discrete model in thecontact areas to continuous model elsewhere shouldbe a solution for future and more realistic studies.REFERENCES1. GDR MIDI, On Dense Granular Flows, Eur. Phys. J. E14, 341-365, 20042. Fillot, Iordanoff, Berthier, “Modelling third body flowswith a discrete element method-a tool for understandingwear with adhesive particles”, Tribology International,Volume 40, Issue 6, June 2007, Pages 973-9813. Godet, M., 1984, “The Third Body Approach: aMechanical View of Wear”, Wear, 100, pp. 437-452.4. Berthier, Y., 1995, “Maurice Godet’s Third Body”, 22 ndLeeds-Lyon Symposium on Tribology, Tribology series,31, pp. 21-30.5. N.Fillot, I. Iordanoff, Y. Berthier : « Wear modelling andthe Third Body concept », Wear, Elsevier, 2007, Vol.262n°7-8, pp. 949-957.6. Iordanoff, I., Charles, J.L., 2007, “Discrete ElementMethod: An Helpful Tool for Abrasion Process Study”,proceedings of the I MECH E Part B, Journal ofEngineering Manufacture, Vol 221, Vol 6, pp 1007-1019.7. Richard, D., Iordanoff, I., Renouf, M., Berthier, Y., 2008,“public ASME 2008”, to be published in ASME Journal oftribology8 Verlet, L., 1967, Computer “Experiments” on ClassicalFluids. I. Thermodynamical Properties of Lennard-JonesMolecules, Phys. Rev., 159, pp. 98-1039 T. Sutatwala, L. Wong, P. Miler, M. D. Feity, J.Menapace, R. Steele, P. Davis and D. Walmer, "Subsurfacemechanical damage distribution during grinding offused silica", J. Non Crys. Sol. 352 (2006) 5601-561710 Vargas, W. L., McCarthy, J. J., 2001, “Heat Conductionin Granular Materials,” AIChE Journal, 47, pp. 1052-10

Simulation of interface temperature and control of lubrication in thestudy of friction and wear in cold rollingK. Louaisil 1 , M. Dubar 1* , R. Deltombe 1 , A. Dubois 1 , L. Dubar 11 Laboratoire d’Automatique, de Mécanique et d’informatique Industrielles et Humaines, UMR 8530Université de Valenciennes et du Hainaut Cambrésis, F-59313 Valenciennes Cedex 9, FranceURL: www.univ-valenciennes.fr/LAMIH/*e-mail: mirentxu.dubar@univ-valenciennes.frABSTRACT: A prototype testing device called upsetting rolling test (URT) was developed a few years agoand was recently optimised in our laboratory to simulate mechanical and thermal contact conditions in coldrolling. Moreover a new experimental protocol has been designed to reproduce the industrial quasi-boundarylubrication regime. Numerous experiments have been then carried out to study the influence of contacttemperature and forward slip on friction, iron fines residues and microscopic surface aspects. A greatinfluence of temperature on friction and wear has been put forward. An increase of the Coulomb frictioncoefficient associated with a decrease of the iron fines quantities has been shown when temperature increaseswhich seems to indicate more adhesive wear.Key words: Cold rolling; Experimental simulation; Lubrication; Interface temperature; Friction; Iron fines1. INTRODUCTIONIn cold rolling, contact at roll-strip interface, whichis the main location of slipping, chattering orimportant wear rate phenomena, has to becontrolled. Thus a simulation test, the UpsettingRolling Test (URT) [1], was developed in ourlaboratory LAMIH to study the contact for theindustrial Sendzimir’s 20-high, a few years ago. Thisreversible mill is used for reduction of low carbonsteel strips. Lubrication is made by an emulsion ofoil in water.All industrial contact conditions have to bereproduced on URT tool. Mechanical, thermal andlubrication contact conditions are first presented andspecified. Then the principle and the methodology ofthe URT and its recent optimisations are described.Finally many tests have been performed in differentcases of contact temperature and relative speed.Results such as mean Coulomb friction coefficient,iron fines pollution and observations of strip surfaceaspects are reported and analysed.2. INDUSTRIAL CONTACT CONDITIONS2.1 Mechanical contact conditionsOperators drive the mill with force parameters: theroll separating force, the rear and front tensions andthe torque. Output parameters are exit speed, V s ,peripheral roll speed, R and reduction rate. Motionparameters are generally defined by the forward slip,S fwd = (V s – R)/R. The strip is driven forwardthanks to the front tension and the frictional forcesbetween rolls and strip [1].2.2 Thermal contact conditionsTemperature has a decisive effect on lubrication,especially on the efficiency of additives [2].The interface temperature depends on a lot offactors: contact partners properties as well as rollingparameters and convective cooling by the emulsionwater are influent. First, heat created by plasticstrain and friction are respectively favoured byreduction rate and rolling speed [3]. Secondly, agood efficiency of adequate lubricant additiveslimits friction and heat generation. Finally, heat

transfer across the roll-strip interface has also adecisive impact. From a general point of view thecontact can be thermally defined by a mean contactthermal resistance which depends on strip and rollproperties, such as roughness and conductivity,contact pressure and lubricant behaviour [4,5].Nevertheless, it is quite important to note that thereare important temperature gradients across theinterface [4]. It present peaks on asperity locationswhere conditions, such as local plastic strain andfriction, are more severe and where a temperature of300°C could be locally reached [6].Therefore, considering all these factors, it is quitedifficult to determine the mean contact temperaturecorresponding to the studied process from literature.Indeed the thermal studies for cold rolling weregenerally carried out on tandem mill with notablythe use of bigger work rolls than on a Sendzimir one[3,4,7]. According to the value of the convectivecoefficient used to define the cooling by emulsion,the mean interface temperature is evaluated between100°C and 160°C [7].As a conclusion, the mean contact temperaturecannot be defined accurately for process analysedherein: a temperature of 120°C will be first tested.2.3 Lubrication contact conditionsis representative of the lubrication regime. In coldrolling process, most of the computations carried outat the end of the roll bite estimate this ratio at morethan 0,8 [8]: strip-roll interface is essentiallygoverned by contact between asperities. Industrialstrips have been observed by SEM after pickling(Fig. 1.a), after two passes (Fig 1.b) and in its finalstate after four passes (Fig. 1.c). Once the secondpass is performed (Fig 1.b), a scaly surface ispointed out: the quasi-boundary lubrication regimeconcerning the studied process is confirmed.3. UPSETTING ROLLING TEST (URT)3.1 ObjectivesThe upsetting rolling test (Fig. 2.) was designed byR. Deltombe & al. [1] and has been recentlyoptimised. The mechanical running of the machineis detailed in [1]. Its aim is to reproduce contactcharacteristics in order to study their influence onthe following parameters:• friction: a mean Coulomb friction coefficient isidentified [1]• iron fines production [1]• surface aspects thanks to SEM observationsFig. 2. Upsetting rolling test designed by authors’ laboratoryLAMIH [1]Fig. 1. SEM observations (a) pickled strip (b) industrial rolledstrip after two passes (c) industrial rolled strip after four passes(d) experimental rolled strip after two passes on URTThe contact ratio, R = A r /A a , where A r and A a arerespectively the real and the apparent contact areas,3.2 Principle: reproduction of industrial contactconditions3.2.a Materials and geometryIndustrial roll and strip specimens are used in orderto reproduce industrial geometry roll bite and realmaterial contact conditions.3.2.b Stresses and plastic strainIt is first important to note that the test is driven in adifferent, but equivalent, way than on industrial mill.The test parameters are sliding forward and

eduction rate (normal and tangential forces aremeasured as output parameters). A methodology wasdeveloped by Deltombe & al. [1]: thanks toindustrial and experimental FEM models, the testparameters enabling the experimental reproductionof industrial contact stresses and industrial plasticstrain are computed.3.2.c Thermal conditionsA convective heating system has been recentlydesigned to control interface temperature. Thissystem is associated with a convective coolingsystem to protect sensors and thus to avoid drift ofmeasurements or even their degradation.3.2.d Lubrication conditionsA new methodology of lubrication application hasbeen designed to solution this problem andreproduce industrial lubrication regime.4. SIMULATION OF LUBRICATION4.1 Lubrication by oil-in-water emulsion in coldrolling processThe Wilson theory, called dynamic concentrationtheory [9] corresponds to the process studied herein.Because of oil high viscosity, this theory supposesthat quasi no water or a tiny amount of water entersthe roll bite [10]: water is mainly used as coolantfluid.The thickness of oil penetrating the contact isdirectly influenced by rolling speeds, oil viscosityand concentration, materials, contact pressure androll bite geometry [8]. In section 2, lubrication wasdefined as almost boundary. Generally the minimallimit of a mixed regime thickness is consideredequal to:h lim = 0,35.R a [6] (1)where R a = mean strip roughness.4.2 Methodology: computation of required oilquantity to applyNo water is considered entering the roll bite: testwill be made with neat oil. Speeds not beingreproducible, lubricant feeding cannot be simulatedon URT: the solution is to apply the good amount ofneat oil on URT specimen beforehand. The usedlubricant thickness for this study is the most criticalvalue we could meet in cold rolling, i.e. h lim , definedin 4.1. The mass corresponding to this thickness iscomputed according to the following procedure.Thanks to a 3D profilometer, the first step is tocalculate the mean roughness, R a , of strip studiedherein and to deduce (1) the corresponding thicknessto apply on it. Then, from a 5 mm² specimen surfacepreviously analysed by the profilometer, computersoftware enables to calculate which volume of oil isnecessary to fill in the valleys in order to obtain therequired lubricant thickness computed in first step.Finally, knowing oil density the mass is deduced.As a validation of these recent optimisations,experimental surface aspect shows a scaly surface asfor the industrial one after two passes (Fig. 1.b, 1.d).5. TESTS AND RESULTS5.1 Test objectives and conditionsThe aim is to analyse the influence of rollingparameters such as the pass number, the interfacetemperature and the forward slip on friction andwear (iron fines creation and surface aspects). Allthe test configurations are indicated for each pass ontable 1. The quantity of lubricant applied onspecimen strip is the mass corresponding to thethickness h lim (Eq. 1). Each configuration of test hasbeen reproduced twenty five times.Table 1. URT test configurations for each passConfiguration number 1 2 3 4Forward slip 2% 2% 7% 7%Interface temperature 40°C 120°C 40°C 120°C5.2 Results and discussions• Friction:The coefficient values corresponding to eachconfiguration are represented on Fig. 3.a:o Coulomb friction coefficient is higher in passtwo than in pass one whatever the testconfiguration. This can be explained by theincrease of the contact ratio with the passnumber. As an evidence, nearly no hole (orlubricant “traps”) remains once pass 2 has beenperformed (Fig 1.b)o Coulomb friction coefficient increases withtemperature. For a quasi-boundary lubrication

egime, the contact behaviour is governed bycontact between asperities where no lubricantfluid film can remain [2]. To prevent realcontacts between roll and strip at asperitieslocations, additives react with surfaces to form aprotective film. The friction reducer additives arethe polar ones and they get desorbed from acritical temperature estimated at about 100°C[2].o No clear influence of forward slip on frictioncoefficient is observed• Iron fines residues:After pass one, iron fines are measured only on roll.After pass two, they are measured both on URTspecimen and roll. The quantities of iron finesresidues collected for each configuration test arerepresented on Fig. 3.b:o The quantity of iron fines measured decreaseswith temperature contrary to friction coefficient.As previously observed (Fig. 3.a), a highercontact temperature causes a higher frictioncoefficient which favours the wrenching ofasperities: a high amount of iron fine residuesshould be found [1] contrary to what has beenobserved and measured (Fig. 3.b). As aconsequence, the high temperature conditionsimply that more wrenched particles must adhere(directly or indirectly) on strip or roll and cannotbe collected.o No clear influence of forward slip on thequantity of iron fine residues is observeda) b)Fig. 3. Results for each test configuration (a) Mean computedCoulomb coefficient (b) Total iron fine residues collected onroll and strip6. CONCLUSIONS AND PERSPECTIVESTo better understand cold rolling contact behaviour anew methodology was created and recentlyoptimised to simulate mechanical, thermal andlubrication contact conditions. It has enabled topoint out that:o friction increases with temperature because ofthe desorption of polar additiveso quantity of iron fines residues decreases withtemperature. It means that adhesive wear andtransfer layer formation is favoured by a hightemperature.o friction increases with the number of the passbecause of the decrease of the contact ratioDespite these interesting conclusions, importantquestions remain: What are the nature and the realimpact of the transfer layer? What is the pertinenceof the current industrial measurement of iron finespollution? Finally, what are the behaviour and therole of the lubricant additives?ACKNOWLEDGEMENTSThe present research work has been supported by Myriad, aCORUS group company, the CNRS, the European Community,the Association Nationale de la Recherche Technique, theConseil Régional du Nord Pas de Calais. The authors gratefullyacknowledge the support of these institutions.REFERENCES[1] R. Deltombe, M. Dubar, A. Dubois, L. Dubar, A newmethodology to analyse iron fines during steel cold rollingprocesses, Wear 254 (2003) 211-221[2] P. Montmitonnet, Tribologie du laminage à froid de tôles,Revue de la Métallurgie Paris, N°2 (2001), 125-130[3] S. Matysiak, S. Konieczny, A. Yevtushenko, Distribution offriction heat during cold rolling of metals by using compositerolls, Numerical Heat Transfer, Part A, 34: 719-729 (1998)[4] A.K. Tieu, P.B. Kosasih, A. Godbole, A thermal analysis ofstrip-rolling in mixed film lubrication with O/W emulsions,Tribology International 39 (2006) 1591-1600[5] A. Boutonnet, Etude de la résistance thermique de contactà l’interface de solides déformables en frottement : applicationaux procédés de forgeage, Thèse INSA de Lyon, 1998[6] J. Molimard, Etude expérimentale du régime delubrification en film mince – Applications aux fluides delaminage, Thèse INSA de Lyon (1999)[7] O. U. Khan, A. Jamal, G. M. Arshed, A. F. M. Arif, S. M.Zubair, Thermal analysis of a cold rolling process – Anumerical approach, Numerical Heat Transfer, Part A, 46:613-632 (2004)[8] N. Letalleur, Influence de la géométrie des aspérités dansun contact hydrodynamique ultramince. Effets locaux etcomportements moyen, Thèse INSA de Lyon (2000)[9] A.D. Bugg, A. Shirizly, J.G. Lenard, The use of emulsionsduring cold rolling of steel strips,, The second Europeanrolling conference rolling , VASTERAS Suède, (2000)[10] K. Dick, J. G. Lenard, The effect of roll roughness andlubricant viscosity on the loads on the mill during cold rollingof steel strips, Materials Processing Technology 168 (2005) 16-24

Metrology of the burr amount - correlation with blanking operationparameters (blanked material – wear of the punch)H. Makich 1 , L. Carpentier 1 , G. Monteil 1 , X. Roizard 1 , J. Chambert 2 , P. Picart 21 LMS – ENSMM, 26 Chemin de l’Epitaphe 25000 Besançon, FranceURL: www.lms.ens2m.fre-mail:hamid.makich@ens2m.fr; Luc.Carpentier@ens2m.fr;guy.monteil@ens2m.fr; xavier.roizard@ens2m.fr2 Institut FEMTO-ST – DMARC, 24 Chemin de l’Epitaphe 25000 Besançon, FranceURL: www.femto-st.fre-mail: jchamber@univ-fcomte.fr; philippe.picart@univ-fcomte.frABSTRACT: Blanking burr and punch wear are two phenomena closely linked. This leads to consider burr asthe best criterion for regrinding or renewing tools. Thus, in order to control the quality and cost in blankingprocess, it is necessary to measure with the best accuracy the amount of burr and wear punch. To do this, amethod for measuring the volume of burr was developed. This method is based on the use of 3Dtopographical images. On the other hand, punch wear was calculated by combination of a mathematicalmodel to the geometric shape of punch. This model gives the geometry variations of the punch, depending onthe number of press storks. At the end, the aim of this study is to identify a relationship between burr andpunch wearKey words: Blanking, Burr amount, Punch wear, Cut edge profile, Worn geometry1 INTRODUCTIONIn the blanking industry, the punch wear has asignificant impact on production. In particular,important wear causes a wrong geometry of the cutpieces, which can lead to a rejection of products.Among other defects, the apparition of burr is themost incapacitating in the use of the blanking piece.So, in order to overcome this problem, industrialsregrind their tools after a number of press storks. Butthis operation has a negative consequence on costs.In fact, the production stopping for regrindingpunches creates substantial losses for the industry.Thus, it is important to control aspects of theblanking operation leading to the appearance of burr.To do this, it is necessary to have in advance areliable way to measure the quantity of burr. In thispaper, we present an original method for thequantification of the volume of burr. Moreover, afirst approximation of the burr amount with the levelof wear and the cutting materiel is also presented.2 BLANKING BURRamount, the cutting pieces can be unusable.Fig. 1.Geometry of the cutting edge.HR : Roll-over depth.HS : Sheared edge.HF : Fracture depth.Hb : Burr height.φ : Fracture angle.Given the important issues related to the appearanceof blanking burr, a lot of work has been undertakento explain the appearance. Most of this work wasdevoted to the study of the influence of blankingconditions (clearance, radius, etc...) (Fig. 2), on theappearance of burr.The burr occurs on the cutting edge of pieces (fig.1),this presence is one of the most serious limitations ofthe blanking process. Indeed, beyond a certain burr

eJ / 2βD pD mr pJ / 2PunchFig. 2.Geometrical parameters characteristic of blanking.The main conclusions of these studies show that thesize of burr depends mainly on the blankingclearance, the wear condition of the tool edge andproperties of the cutting materials and the toolmateriel. According to the studies carried out in [1]and [2], it appears that the wear promotes theappearance of burr and that the increase in theblanking clearance results in an increase in thefracture angle of the shear zone and the height of theburr. This is confirmed also in [3] and [4]. Withregards to the properties of blanking materials,Gréban in [5] says that the composition of cuttingalloys has a strong influence on the quantity of burrsthat is created during the blanking operation.However, the works are more on the theoreticalstudy of the conditions of burr appearance, than onan experimental metrology on the burr amount.αr mBlank-holderSheet metalDie3.1 MetrologyWe can see that in literature, there are many ways ofmeasuring burr, but fundamentally based on adestructive test of blanking pieces and therefore ofmetrology in a few positions of the cutting edge.Other methods that have been proposed formetrology burr heights were the non-contact ortactile optical profilometry [7], or the visionmetrology [8] of the cutting edge, which is aimed atpieces with straight faces. The latter allows thenaccess to the profile heights of burr over a length ofcut board of 4mm. Another example is ameasurement using the shadow of burr or a moretraditional measure, the mechanical profilometry.3.2 Crushed burrIn the case of progressive blanking, complexgeometry of the parts is produced by several pressstrokes, which has the effect of crushing burrs. Thismakes it more difficult to measure the height of theburr by traditional means.3 METROLOGY OF BURRThe possibility to estimate the burr size on the blankpieces remains of great importance in the blankingindustry, because the quality of the products isdetermined with the evaluation of the level ofacceptable burr on the parts. It is possible to study,in theory, the emergence of blanking burrs.However, if we wish to verify the theoreticalpredictions, it is necessary to be able to accuratelyand reliably measure the quantity of burrs. Tosummarize, we can return to what might be adefinition of the burr height, the main criterionmentioned here. According to [6], because the sheetcan have a residual macroscopic deformation duringthe cutting operation, the height of burr is defined as"the difference between the highest point of burr andthe surface of the sheet metal immediately adjacentto the burr". This definition seems satisfactory andwill be the basis for development of our work.Fig. 3. Crushed burr.This has prompted us to seek, in this work, a newmethod more able to give us a precise quantificationof the crushed burr. To do this, we sought to define areference plan around the cutting form, because theadjacent sheet is distorted.3.3 Establishing a means of measurementIn our approach to the burr quantification, we soughtto meet a need, which is to confront the theoreticalresults with experimental robust measurements.3.3.a Images AcquisitionThe device used is a microscope Infinitefocus ®

Alicona (Fig.4). It is based on an optical microscopecoupled to a camera. Unlike a confocal microscope,which uses a monochromatic sensor, the device usesa sensor of contrast in color. On the whole height,the microscope collects images which are thenanalysed by the software. The latter analyses eachpixel and compares it to its neighbours to see if it isfocused or not. Then, it reconstructs a threedimensionalimage from this focus. From this image,surface-, volume-, surface state- and topographydimensional measurements are possible. Inaccordance with selected lens, the image size rangesfrom 2.1x1.6mm ² to 103x83 µm² and resolution inheight from 444nm to 20nm. If the sample or area ofinterest is larger than the size of capture of theselected lens, it is possible to achieve a multi-fieldacquisition through motorized tables. We only needto identify the entire area to be analysed, and byinterconnection between image fields acquired withrecoveries, we reconstructed a joined-fie image.Profile afterstraighteningProfile beforestraighteningFig. 5 . Profiles on the surface of the sheet before and afterstraightening .• Secondly, an image processing is performed inorder to extract the profiles that we wishvertically from the contour of the blanked shape(Fig. 6).Blanked ShapeSheet metalProfile beginningProfileFig. 4. The optical device measurement: InfiniteFocus ®3.3.b MeasurementThe measurement method is developed within theLMS specifically for quantifying the burr volume.The originality of this method lies in thetransformation of the blanked shape, (here a circularshape). The main stages of this transformation are:Fig. 6. Example of obtained profile.• Thirdly, the profiles are extracted in multiples soas to build a rectangular matrix; each profilebeing the average of 100 adjacent profiles.Thereafter, the image is adjusted, outside thezone of the blunder, with a linear polynomial ofdegree 2, which assembles all the profile (Fig. 7).• First, a straightening of the image around theblanked shape (hole) is achieved (Fig. 5). This isdone in order to eliminate any imperfections inthe image rectitude. The straightening of theimage (outside the zone of burr) is made with themethod of least squares.Fig. 7. Representation of the profiles assembling.• Finally, the volume is determined only in theareas of the burr presence on the image.

number of profiles, which has the advantage ofallowing a complete scan of the blanking edge.REFERENCESFig. 8. Calculations on the defined area4 RESULTS AND CONCLUSIONSFigure 9 shows the evolution of the burr volumeaccording to the number of press stroke. We observean increase of a burr volume with the increasingnumber of press stroke.Burr volume(µm3)1,20E+061,00E+068,00E+056,00E+054,00E+052,00E+050,00E+000 50000 100000 150000 200000 250000 300000Number of press strokesFig. 9. Burr volume.Other results (Fig. 10) can show several veryinteresting phenomena. Indeed, a quick initialanalysis of the curve allows noting that the kineticsof appearance of burr, as it was highlighted byGréban [5], depends directly on the blankingmaterial.1. N. Hatanaka, K. Yamaguchi, N. Takakura, T. Iizuka «Simulation of sheared edge formation process inblanking of sheet metals » Journal of MaterialsProcessing Technology, 140 (2003) 628-634.2. C. Husson, C. Poizat, L. Daridon, S. Ahzi « Travail desmétaux en feuilles - Simulation numérique 2D dudécoupage d’alliages de cuivre » 7ème ColloqueNational en Calcul des Structures, 2005.3. A. Touache « Contribution à la caractérisation et à lamodélisation de l’influence de la vitesse et de latempérature sur le comportement en découpage de tôlesminces. » Thèse à l’Université de FRANCHE-COMTÉ,14 Décembre 2006.4. Z. Tekiner, M. Nalbant, H. Gürün « An experimentalstudy for the effect of different clearances on burr,smooth-sheared and blanking force on aluminium sheetmetal » Materials & Design, 27 (2006) 1134-1138.5. F. Gréban, G. Monteil, X. Roizard « Influence of thestructure of blanked materials upon the blanking qualityof copper alloys » Journal of Materials ProcessingTechnology, Vol 186 N°1-3, (2007), pp 27-32.6. M. Grünbaum, J. Breitling, Influence of high cuttingspeeds on the quality of blanked parts. ERC report N°5-96-19, 1996.7. F. Gréban « Découpabilité du cuivre et des alliagescuivreux. » Thèse à l’Université de FRANCHE-COMTÉ, 21 février 2006.8. E. Levy « Mise en oeuvre des systèmes de vision pour lecontrôle des pièces découpées ».Document interneStatimage. 2000.Burr volume(µm3)1,60E+061,40E+061,20E+061,00E+068,00E+056,00E+054,00E+052,00E+05Mat 1Mat 2Mat 1 Mat 10,00E+000 100000 200000 300000 400000 500000 600000 700000 800000Number of press strokeMat 3Fig. 10. Burr volume for different materials.Because of the non-homogeneity of the burrs on theblanked contour, the method seems the mostappropriate, because it is obtained from a large

Laboratory and field analysis on the tribological behavior of coated anduncoated forming toolsM.A.R.S. Mendes 1 , R.M. Souza 1 , P.K. Vencovsky 2 , Y. Berthier 31 Surface Phenomena Laboratory, Department of Mechanical Engineering, Polytechnic School of theUniversity of São Paulo – Av. Prof. Mello Moraes, 2231, 05508-900 São Paulo, SP, BrazilURL: www.poli.usp.bre-mail: roberto.souza@poli.usp.br; marcorsm@usp.br2 Bodycote Brasimet – Av. Nações Unidas, 21476, 04795-912 São Paulo, SP, BrazilURL: www.brasimet.com.bre-mail:Paulo.vencovsky@bodycote.com3 INSA de Lyon, LaMCoS – 20 avenue Einstein, 69621 Villeurbanne, FranceURL: lamcos.insa-lyon.fre-mail: yves.berthier@insa-lyon.frABSTRACT: Frequently, the life of the tools used in sheet metal forming operations is determined by aphenomenon known as galling, which originates from the adhesion of the sheet to the forming tool surface.The application of coating architectures composed by single or multiple layers of Physical Vapor Deposition(PVD) films, such as TiN, TiCN, CrN, TiCNAl, may significantly reduce the chemical interaction in thecontact, up to the point that no significant adhesion may be observed for an extended number of formingoperations. Usually, the evaluation of the behavior of different thin film architectures is conducted usingtribometers that may or may not reproduce the conditions found in industrial practice. This work presents atribological analysis of coated and uncoated surfaces of tools used in industrial sheet metal formingoperations and discusses the capability of laboratory tests in reproducing the situations found in practice.Key words: Galling, PVD coatings, Laboratory tests, Field tests1 INTRODUCTIONFrequently, the life of the tools used in sheet metalforming operations is determined by a phenomenonknown as galling. Galling is described in the ASTMG40 Standard [1] as “a form of surface damagearising between sliding solids, distinguished bymacroscopic, usually localized, roughening andcreation of protrusions above the original surface; itoften includes plastic flow or material transfer orboth”. In spite of the significant number ofpublications on this phenomenon, some attention isstill devoted towards its effective occurrence [2]and, more importantly, towards the development oftests able to reproduce galling in laboratory [2-10].One important reference regarding these tests is theASTM G98 method [3], according to which athreshold pressure for galling is calculated based onthe visual inspection of the presence of galling at thesurfaces of a button and/or a block that were pressedagainst and manually rotated with respect to eachother at increasing loads. Despite the popularity ofthe ASTM G98 test method, many works criticizethe applicability of such threshold pressure value fordesign purposes and suggest modifications to theoriginal standard method. These changes intend toovercome some of the method limitations, such asthe heterogeneity of contact pressure distribution [4]or the fact that the velocity is zero at the centre ofthe rotating button [2,4]. Other questionings on theASTM G98 test method include the nonconsideration of the statistical nature of galling[4,5]; the absence of constant speed during manualrotation [6] and even cost issues associated with thenecessity of a large number of specimens or theavailability of an equipment capable of applyinglarge normal loads [7].Some of the alternatives for laboratory testing ofgalling include a modification in the geometry of thespecimens, such as the shape of initial contact,which is considered as a line [8] or a point [7,9],rather than an area (ASTM standard). In these tests,the analysis of the occurrence of galling may remainbased on the visual inspection of contact surfaces [8]or, as in the case of two crossed cylinders that slideagainst each other, may be based on the frictioncoefficient calculated during the experiments [7,11].

In spite of the predominance, galling is not the onlyphenomenon observed in forming operations and,during the past decades, more sophisticated testshave been developed [10,12-15] aiming a betterlaboratory reproduction of the conditions found inpractice. Most of these tests are conducted inspecially designed rigs and involve the actualcontact of metal strips against tool materials.Many of the laboratory studies mentioned above[2,9,13] have also explored the point that theevolution of galling at a given surface presents acorrespondence with variations in roughness [16].The intensity and continuity of the search for alaboratory test that accurately reproduces thetribological conditions in forming operations areunderstandable, not only based on the possibility ofknowing these operations in detail, but also based onthe ability that such test would have to provide hintstowards materials development. In particular, suchtest would provide a means of keeping track of theapplicability in forming operations of constantlyemerging developments of Physical VaporDeposition (PVD) coating architectures.In this work, eight sheet metal operations, conductedwith coated and uncoated tools, were analysed interms of wear phenomena observed at tool surface.Results were then discussed in the light of thelaboratory tests mentioned in the previousparagraphs.2 FIELD TESTS AND ANALYSISTable 1 presents a list of the industrial sheet metaloperations studied in this work. All operations werepart of the actual production line of a bearingindustry. Operations 1 to 6 were conducted as stepsof the manufacturing process of cups with diameterof 30 mm and depth of 25 mm. Similarly, operations7 and 8 were part of the manufacturing procedure ofcups with diameter of 16 mm and depth of 12 mm.As indicated in the table, two types of tool steel wereselected for the tools (or substrates), AISI H13 andAISI M2, which presented Rockwell C hardness of52 (approx. 6.0 GPa) and 61 (approx. 7.5 GPa),respectively. The TiCN and TiCNAl coatings weredeposited in a commercial cathodic arc evaporationdeposition chamber and, according to themanufacturer, presented hardness on the order of 29GPa and 35 GPa, respectively. All operations listedin table 1 were lubricated with oil and conducted on0.8 mm thick 16MnCr6 steel sheets. In table 1, thecolumn labelled as “Production” indicates thenumber of parts manufactured by each tool, whoselife limiting factor was based on the quality of theformed part. The end life of all uncoated operations(1-4) was determined by galling and the end life ofall coated tools was not determined by galling, butby other wear phenomena that resulted from thecontact between the punch and the die (e.g. wear onthe sides of the punches due to slight centremisalignment).Table1. Overall conditions in sheet metal forming operations.Operations were lubricated with oil and conducted on 8.0 mmthick 16MnCr6 steel sheetsOperationnumberSubstrate Coating Type ofOperation1 AISI H13 None Ironing 322 AISI M2 None Blanking 1753 AISI M2 None Bending 1754 AISI M2 None Coining 945 AISI H13 TiCN Ironing 10006 AISI M2 TiCN Coining 10007 AISI M2 TiCNAl Ironing 12008 AISI M2 TiCNAl Coining 1200Production(x10 3 parts)After removal from the press, tool surfaces werecleaned and observed with optical and scanningelectron microscopy (SEM). Surfaces were alsoanalysed in terms of roughness.Figure 1 presents a detail of the surface of the toolused in operation 2. Two distinct regions may beobserved in figure 1, one that presents the originalsurface, and an external ring where the originalgrinding lines are no longer visible. Additionalanalyses of the external ring revealed featurestypical of galling, such as surface protrusions(figure 2) and the presence of additional peaks andvalleys in the roughness profile (figure 3).Fig. 1. Optical microscopy of the tool used in the blanking of8.0 mm thick 16MnCr6 steel sheets (operation 2 in table 1)

as an absence of particles adhered to the tool surfaceand the absence of additional peaks and valleys inthe roughness profile.a50 µmFig. 2. SEM analysis of the tool used in the blanking of 8.0 mmthick 16MnCr6 steel sheets (operation 2 in table 1). Arrowscoincide with the radial direction of the cylindrical tool: (a)Secondary electron image and (b) backscattered electron imagebUnwornWorn ringµmammFig. 3. Roughness analysis of the tool used in the blanking of8.0 mm thick 16MnCr6 steel sheets (operation 2 in table 1): (a)Unworn surface and (b) worn external ring.Operations 4 and 6 are representative of thedifferences observed with coated and uncoated tools,since the same tool geometry and overall operationconditions were present in both cases. In the toolused in operation 4, it was once again possible toobserve that worn surfaces present additional peaksand valleys in the roughness profile, which were notclearly noticeable at the surface of tool used inoperation 6.Differences between coated and uncoated surfaceswere also observed in the SEM analysis. Although ina scale different from that shown in figure 2, gallingfeatures were also observed at the surface of theuncoated tool used in operation 4. These featureswere not distinguishable at the TiCN coated surface(figure 4), which presented several scratches in theradial direction, but where it was not possible toidentify the presence of material transferred from thesheet to the coated tool surface, in spite of the largernumber of parts produced with this tool.3 DISCUSSIONµmAs a general trend, coated tools, independent of thecoating material, outperformed uncoated tools in allindustrial operations analysed in this work. Thebetter performance of the coated tools may beattributed to their ability to minimize, or postpone,the occurrence of galling; an ability that was verifiedbmmFig. 4. SEM analysis of the TiCN coated tool used in thecoining of 8.0 mm thick 16MnCr6 steel sheets (operation 6in table 1).It is not appropriate to make a direct comparison ofthe results obtained in this work with those of thetests presented in the literature, since most of theelements of the tribossystem were different,including materials, geometries, loads andlubrication condition. However, some considerationsare still possible and, to this end, the laboratory testswere intentionally grouped into three categories,which are the tests based on the ASTM G98standard; tests with initial line or point contact andtests that involve actual sliding of strips against thesurface of tools.It is possible to suppose that the use of the toolmaterials of this work in tests based on the ASTMG98 standard would provide lower threshold gallingpressures for the uncoated tools than the coatedones. However, in addition to the questionings onthe concepts of ASTM G98 standard [2,4-8],literature [4,9] data also indicates that thepreparation of these tests must be careful in order toprovide good alignment between the contactingsurfaces, even if some of the alternatives to theoriginal standard are selected.The simplicity and quickness of tests with line orpoint contact may be advantageous. However, takingthe example of the crossed cylinder test, it ispossible to state that the use of the outputs of thistest to study forming operations is based on theassumptions that: (i) galling is associated with anincrease in friction coefficient and (ii) galling is thepredominant mechanism in forming operations. Thefirst of these assumptions may seem intuitivelycorrect and a direct correlation between friction andgalling has been observed in some cases [14]. On the

other hand, it has not been observed in others [2].The second assumption is also plausible, but theliterature also presents examples where wear waspredominantly dictated by ploughing, rather than bygalling [14]. In terms of the comparisons providedby the crossed cylinder test, Podgornik [7] observedthat uncoated tools performed better than toolscoated with titanium nitride (TiN). Based on theresults of the industrial operations analysed in thiswork, it is difficult to imagine that this result wouldoccur in industry. Therefore, in addition to thevalidity of the assumptions presented above, it ispossible to suppose that some of the conditionsimposed during the line or point contact tests areunable to entirely reproduce the conditions found inpractice. One of the possible explanations for thesedifferences is the nominal contact pressure, which,in these tests, tends to be larger than in the industrialconditions [9]. A reduction in pressure would bepossible in the crossed cylinder test, but, in this case,the sliding distance would probably have to besignificantly increased until detectable wear wouldbe noticed. As a result, repetition of the regions incontact would possibly occur during each test cycle;a condition that is frequently not observed informing operations [12].Finally, it is possible to expect that tests that involveactual sliding of strips against the surface of toolswould provide results in good agreement with thoseobserved in industrial operations. However, thesetests frequently demand specially designedequipment and may require large and complexshaped specimens.4 CONCLUSIONSThe literature presents a large number of laboratorytests dedicated to the study of the wear phenomenonknown as galling and/or the tribology in formingoperations as a whole. A qualitative comparison ofsome of these tests with results obtained directly inindustrial operations indicate that, in spite of thelarge number, most of the options are still notentirely satisfactory. Apparently, none of them isable to provide a simple means to analyse factorssuch as contact pressure, lubrication, continuous orintermittent contact, refreshment of contact area, inorder to not only analyse galling, but also, ifnecessary, conduct the test with different levels ofpredominance of galling over other phenomena.ACKNOWLEDGEMENTSThe authors acknowledge two Brazilian founding agencies forfinancial support: the São Paulo State Research Foundation -FAPESP through projects number 2006/02006-2 and2003/10157-2 and The National Council for Scientific andTechnological Development (CNPq) through project550235/03-5.REFERENCES1. ASTM G40, Standard Terminology Relating to Wearand Erosion, ASTM International, West Conshohocken(2005).2. K.G. Budinski, M.K. Budinski and M.S. Kohler, Agalling-resistant substitute for silicon nickel, Wear 255(2003) 489-97.3. ASTM G98, Standard Test Method for GallingResistance of Materials, ASTM International, WestConshohocken (2002).4. S.R. Hummel, Development of a galling resistance testmethod with a uniform stress distribution, Tribol. Int. 41(2008) 175-80.5. S.R. Hummel and B. Partlow, Comparison of thresholdgalling results from two testing methods, Tribol. Int. 37(2004) 291-5.6. K. Gurumoorthy, M. Kamaraj, K. Prasad Rao and S.Venugopal, Development and use of combined weartesting equipment for evaluating galling and high stresssliding wear behaviour, Mater. Des. 28 (2007) 987-92.7. B. Podgornik, S. Hogmark and J. Pezdirnik, Comparisonbetween different test methods for evaluation of gallingproperties of surface engineered tool surfaces, Wear 257(2004) 843-51.8. S.R. Hummel, New test method and apparatus formeasuring galling resistance, Tribol. Int. 34 (2001)593-7.9. P.A. Swanson, L.K. Ives, E.P. Whitenton and M.B.Peterson, A study of the galling of two steels using twotest methods, Wear 122 (1988) 207-23.10. L.M. Bernick, R.R. Hilsen and C.L. Wandrei,Development of quantitative sheet galling test, Wear 48(1978) 323-46.11. M. Hanson, N. Stavlid, E. Coronel and S. Hogmark, Onadhesion and metal transfer in sliding contact betweenTiN and austenitic stainless steel, Wear in press.12. E. Schedin, Galling mechanisms in sheet formingoperations, Wear 179 (1994) 123-8.13. F. Clarysse, W. Lauwerens and M. Vermeulen,Tribological properties of PVD tool coatings in formingoperations of steel sheet, Wear 264 (2008) 400-4.14. C. Boher, D. Attaf, L. Penazzi and C. Levaillant, Wearbehaviour on the radius portion of a die in deep-drawing:Identification, localisation and evolution of the surfacedamage, Wear 259 (2005) 1097-108.15. M. Hirasaka and H. Nishimura, Effects of the surfacemicro-geometry of steel sheets on galling behavior, J.Mater. Process. Technol. 47 (1994) 153-66.16. J.L. Andreasen, N. Bay and L. de Chiffre, Quantificationof galling in sheet metal forming by surface topographycharacterization, Int. J. Mach. Tool Manu., 38 (1998)503-10.

DFT – Modeling of the Reaction of a Polysulfur Extreme-PressureLubricant Additive on Iron SurfaceB. Monasse, P. MontmitonnetEcole des Mines de Paris PARISTECH, CEMEF, UMR 7635, BP 207, 06904 Sophia-Antipolis Cedex, FranceURL: www-cemef.cma.fr e-mail: bernard.monasse@ensmp.fr e-mail: pierre.montmitonnet@ensmp.frABSTRACT: sulfur-containing molecules are used as extreme pressure additive for low-alloy steel surfaces.These molecules react with ferrous surfaces and form FeS x compounds which strength the surface and reducescuffing. The stable conformation and the chemical stability and reactivity of an alkyl sulfur molecule(DTDP) with oxygen and an iron surface are here predicted by DFT model. The effect of the conformation ofthe molecule respective to the surface acts on the surface reactivity.Key words: polysulfur, iron, reactivity, semi-empirical functional,1 INTRODUCTIONSulfur-containing organic molecules have been usedfor a long time as extreme pressure additives inlubricating oils. Several papers report oncomparisons of the tribological efficiency of organosulfuradditives as a function of number of sulfuratoms, alkyl- or aryl-chains [1]. In tribotests,involving ferrous materials and S-containingadditives, the formation of the FeS x compound hasbeen detected in superficial layers by differentsurface analysis techniques [2, 3]. FeS, also knownas troilite, has a layered hexagonal structure, lowshear strength and high melting point (1100 °C), soit is an effective solid lubricant like graphite andMoS 2 [4]. HSAB theory may also shed light on thosespecies which react to form these layers [5]. Themolecule may be thermally decomposed or oxidizedin the lubricant during rolling process or reacts onthe metallic surface. The details of the reaction pathare nevertheless poorly known, and so is the effectof the environment (dissolved oxygen, temperature,…) on preferred reaction paths and reactivity andwill be analyzed in the present work.Quantum models have been applied to study thereactivity of some sulfur molecules on variousmetallic surfaces [6]. Most of the papers are focusedon copper, palladium and nickel for catalyticpurpose with simple sulfur molecules. Reactions acton the first layers of metal mainly for gas phasereactions [6,7]. The sulfur may be randomlydistributed on copper surface or organized asmobile sulfur metal clusters Cu 3 S 3 [8]. Theorganisation of metal sulfur molecules may dependon the reaction path of the sulfur during its reactionwith metal surface. We consider here the reaction ofa di-tertio-dodecyl-pentasulfur molecule (C 12 H 25 S 5C 12 H 25 , DTDP) with a bcc iron [001] surface.2 SIMULATIONThe simulations have been done with a semiempiricalmodel MOPAC using the functional PM5.The results were confirmed by simulations withother semi-empirical functionals AM1 and PM3 anda DFT functional B88-LYP. The MOPAC softwareis coupled to the graphical user interface CAChe.Full geometry optimization of a standalone moleculeis searched first to define the equilibriumconformation of the molecule. This conformation issearched from various initial non-equilibrated state.A molecular dynamics simulation allows us tofollow the conformational change around theequilibrium conformation. The stable conformation

corresponds to the minimum energy state. Thereactivity of this molecule at the contact is thenpredicted with various approaches of this moleculenear the metal surface.3 RESULTS3.1 Molecular conformation of DTDP moleculeThe DTDP molecule is symmetrical from the centralsulfur atom, which may act of its equilibriumconformation. A fully extended molecule is createdwith CAChe software and relaxed with varioussemi-empirical and DFT functionals at 0K (fig. 1).C1C2C3C4C5C6C7C8C9C10C11C12C29C28C27C26C25C24C23C22C21C20C19C18S13 S14 S16 S17S15a b cFig. 1. Equilibrium conformation of DTDP molecule along thethree main orientation directionsThe molecule forms a loop with parallel alkyl chainsin extended conformation. The loop results fromgauche conformations of S-S bonds. The parallelchains results from maximal van der Waals andelectrostatic interactions of aliphatic segments. Theconformation change of the molecule results onlyfrom the S-S bond rotation. The aliphatic chains arealmost normal to the mean plane containing sulfuratoms. Complementary simulations with a variablenumber of sulfur atoms show that this loopconformation appears from three sulfur atoms up.Molecular dynamics simulations at 300K of a DTDPmolecule show out a fluctuation of orientation anddistance of almost all-trans aliphatic sequencearound the equilibrium conformation. They comefrom fluctuations of dihedral angle S-S-S around theequilibrium conformation. The loop corresponds tothe equilibrium conformation of the molecule. Theconformational energy decreases by 57.3 kJ.mol -1from the fully extended molecule to the stableconformation.3.2 Stability of bonds and oxidationThe homolytic scission of S-S bonds and theiroxidation are two main hypotheses proposed toexplain the reactivity of sulfur based molecules. Wehave calculated the intrinsic stability of the variousbonds (S-S, C-S, C-C) inside the molecule and theirreactivity with O 2 molecules.The energy of each bond is estimated by thefollowing energy balance deduced from calculationof the energy of the molecule and of the molecularfragments resulting from the bond leakage:E bond = E fragment1 + E fragment2 - E molecule (1)The atoms are numbered from one methyl carbonand sulfurs range from S13 to S17, and each bond isdefined by the linked atoms (fig. 1). The bondenergy depends on the nature and localization ofbonds inside the molecule, half molecule isrepresented thanks to molecular symmetry and mostof C-C bonds are not considered for their high bondenergy (table 1).Table 1: bond energy inside DTDP moleculebond Energy kJ.mol -1C4-C5 268C11-C12 246C12-S13 206S13-S14 215S14-S15 164The sulfur and C-S bonds are significantly weakerthan carbon bonds and the two bonds linked to thecentral sulfur are the weakest leading to the mostprobable position for an homolytic rupture.The following mechanism is considered for athermo-oxidative degradation (fig. 3) :C 12 H 25 S 5 C 12 H 25 + O 2 → R-S-O . + R’-S-O (2)and leads to the following reaction energy (table 2)Table 2: energy of oxidation reaction on varioussulfur bondsreaction Energy (kJ.mol -1 )C12-O S13-O - 83S13-O S14-O - 171S14-O S15-O -337

The high oxidation energy on bond S14-S15 maysuspect a thermal oxidation of the molecule.The relaxation of this molecule on the iron surfacepredicts the reaction of two non consecutive sulfuratoms with iron atoms (fig. 4).Fig. 2. excited state for oxidative reaction on S14 and S15atomsThe bond S14-S15 is the weakest bond for oxidationand for homolytic scission. This part of the moleculeis a clear candidate to react with a metallic surface.3.3 Reaction of DTDP molecule with iron surfaceA direct reaction on an iron surface is also analyzed.A layer of Fe atoms is created and locked torepresent a surface of bcc α-iron, with the samecrystallographic parameters. The iron functional isavailable inside PM5 basis-set and is able to predictreaction with DTDP sulfur based molecule. A DTDPmolecule is then located near the surface withvarious orientations of sulfur atoms. Figure 3presents one initial scheme.Fig. 4. Conformation of a DTDP molecule after reaction withthe iron surface (tilt/iron plane ~ 60°)This type of reaction leads to isolated sulfur-ironbonds distributed on the surface which are weakenedinside the DTDP molecule (fig. 6).Other simulations with the same molecule but with adifferent tilt-angle with the surface lead a differentreaction path (fig. 5).Fig. 5. Conformation of a DTDP molecule after reaction withthe iron surface (side and bottom view) (tilt/plane ~ 80°)Fig. 3. An initial conformation of a DTDP molecule near theiron surfaceAll the sulfur atoms react with an iron atom, ormore, which leads to a cluster of Fe-S bonds on ironsurface. These bonds are strong and weaken theintramolecular bonds, measured by the reducingtheir covalent order (fig. 7).

10.90.80.70.60.50.40.30.20.100 5 10 15 20 25 30Fig. 6. Covalent order of intramolecular bonds inside DTDPmolecule (from fig. 4)10.90.80.70.60.50.40.30.2bond orderordre liaisonbond orderC12-S13S13-S14S14-S15S13-S14S14-S15S16-S17S16-S17S17-C18S17-C18S15-S16bond0.1bond liaison00 5 10 15 20 25 30Fig. 7. Covalent order of intramolecular bonds inside DTDPmolecule (from fig. 5)The two (fig. 6) and four S-S bonds orders (fig. 7),respectively, are almost reduced to none afterreaction on iron surface in favor to Fe-S bonds. Theiron surface is highly reactive to sulfur atoms. Theresulting sulfur groups can be directly organized incluster and randomly distributed as a function on theapproach path of the DTDP molecule onto the ironsurface.4 CONCLUSIONSThe conformation and the stability of the bonds of asulfur based molecule are mainly dependent onsulfur bonds. These ones are much more reactiveand oxidized than carbonyl bonds. Sulfur atomspreferentially react with iron surface and lead torandomly distribution of isolated Fe-S bonds and onsmall clusters of Fe-S as an effect of contact path ofthe molecule on the surface. These results, in onehand, should be completed with similar analyses onother crystallographic planes of iron and of ironoxide surfaces and, in other hand, by prediction ofthe reaction rate of oxidation or bond rupture bycalculation of the excited state energy.REFERENCES1. K.D. Allum and J.F. Ford, The influence of chemicalstructure on the load carrying properties of certainorgano-sulfur compounds, J. Instit. Petrol., 497 (1965)145-169.2. I. M. Petrushina, E. Christensen, R. S. Bergqvist, P. B.Møller, N. J. Bjerrum, J. Høj, G. Kann and I.Chorkendorff, On the chemical nature of boundarylubrication of stainless steel by chlorine- and sulfurcontainingEP-additives, Wear, 246 (2000) 98-105.3. G. Dauchot, R. Combarieu, P. Montmitonnet, M.Repoux, G. Dessalces and F. Delamare, Tribochemicalreactions in cold-rolling : a ToF-SIMS study of thechemisorption of the lubricant additives on the sheet,Rev. Met. Paris, CIT-Science et Génie des Matériaux, 2(2001) 159-168.4. D.M. Zhuang, Y.R. Liu, J.J. Liu, X.D. Fang, M.X.Guang and Y. Cui, Microstructure and tribologicalproperties of sulphide coating produced by ionsulfuration, Wear, 225 (1999) 799-805.5. R.G. Pearson, Chemical hardness, Wiley-VCH, New-York (1997).6. H. Toulhoat, P. Raybaud, S. Kaztelan, G. Kresse, J.Hafner, Transition metals to sulfur binding energiesrelationship to catalytic activities in HDS: back toSabatier with first principle calculations, Catal. Today,50 (1999) 629-636.7. D.R. Alfonso, First-principles studies of the √7 x √719.1° structure of sulfur on the Pd(111) surface, SurfaceScience, 601 (2007) 4899-4909.8. G.B.D. Rousseau, A. Mulligan, N. Bovet, M. Adam, V.Dhanak and M. Kadodwala, A structural study ofdisordered sulfur overlayers on Cu(111), SurfaceScience, 600 (2006) 873-903.

Research on Friction during Hot Deformation of Al-Alloys at HighStrain RateP. Petrov 1 , M. Petrov 1 , E. Vasileva 1 , A. Dubinchin 11 Moscow State Technical University “MAMI” – B.Semenovskaya str. 38, 107023 Moscow RussiaURL: www.mami.rue-mail: petrov_p@mail.ru; petrovpa@online.ruABSTRACT: The paper is linked to the investigation of lubricants for massive hot forging. It was observedtemperature range of 200-470°C and screw press was used for tests. The research on friction has been donefor Al-Mg and Al-Cu-Mg-Fe-Ni aluminium alloys. In this connection, physical and numerical investigation offriction are performed. On the basis of the ring upsetting technique, the effect of temperature on the frictionfactor value has been investigated. The regressions for the relationship between friction factor andtemperature for all lubricants under study have been obtained. Moreover, the sets of calibration curves weredrawn. Each set of calibration curves corresponds to the definite type of aluminium alloy as well as definiteconditions of deformation. Some practical recommendations were given.Key words: interfacial friction, friction factor, aluminium alloy, non-ferrous alloy, isothermal hot forging,lubricant, massive hot forming1 INTRODUCTIONMassive hot forging of any aluminium alloy can beperformed with the help of several types of metalformingmachines, namely hydraulic press,mechanical press, screw press etc. Applying one ofthese machines allows production of either typicalforgings or near net shape forgings. On the otherhand, the quality of a forging made of an Al-alloydepends on the composition of a lubricantsignificantly. The choice of the lubricant for hotforging is major task, especially in case ofaluminium alloys deformation. Moreover, inaccuratedescription of friction for numerical simulation of aforging technology may be one of the reasons whythe results of simulation and experiments are not in agood agreement.The tribological properties of the lubricant can bedetermined with the help of the ring-compressiontest which is the most simple and widely usedmethod for the quantitative estimation of the contactfriction during bulk metal forging. It was developedby Kunogi [1], Male & Cocroft [2] and furthermodified by Burgdorf [3] etc.This method allows to define a friction factor’svalue for the definite temperature only. But thenumerical analysis of forging technology requiresknowing the effect of temperature or/and strain rateon the value friction factor.In connection with it as well as hot forging of Alalloysthere is a lack of systematized dataconcerning the research on this effect for wide rangeof temperatures of forging and strain rates. Forinstance, in [4] it is only presented the results ofeffect of temperature on friction factor duringdeformation of Al-5Si and Al-4Mg alloys. Thesedata were obtained for temperature range of 20-500°C but it is unknown what the value of strain ratein those tests was. Then in [5] the properties of twodifferent lubricants were investigated duringdeformation of Al-Mn, Al-Mg, Al-Cu-Mg and Al-Cu-Mg-Fe-Ni alloys within temperature range of200-470°C. In this case the hydraulic press was usedfor deformation. It provided the initial strain rate of0.14 s –1 . But the question is whether the effect oftemperature will be significant if the screw press ischosen for deformation? It implies that thetemperature range and the materials under studyshould be the same as in [5].So, the aim of the present paper is linked to theinvestigation of effect of temperature on frictionduring the deformation of Al-alloys on the screwpress and farther generalization of the obtained datawhich can be applied in industry in order to optimizethe new or/and being technological processes ofmetal forming. In the present paper, the ring test anda numerical simulation were applied to estimate thetribological properties of lubricants.

2 EXPERIMENTAL PROCEDURETwo lubricants were chosen for the investigation asin [5]. These are the lubricant based on industrial oil(IO+G) and based on synthetic oil (SO+G). Bothlubricants contain colloidal graphite particles aslubricant’s components and the size of particles isless than 15 µm. The chemical composition of alloysunder study is given in Table 1 [5]. The bold type inTable 1 indicates the amount of the basic impurities,which the investigated alloys contain.The sizes of the ring samples were as follows: innerdiameter = 20 mm; outer diameter = 40 mm; height= 14 mm. The ring samples were heated totemperatures of 200, 350, 430, 450, and 470°C in anelectric furnace. Deformation of the heated sampleswas carried out on flat dies that were warmed up totemperature of 120°C. Samples were compressedwith lubrication. Die velocity was constant at V≈400mm/s (screw press with nominal load = 10 MN andmaximum load = 16 MN), which corresponded to aninitial strain rate of 28.6 s –1 . This value of strain ratebelongs to the strain rate interval (10 0 -10 2 s -1 ) withinthat the massive hot forging is usually carried out.Table 1. Chemical composition of alloys [5]Element Percentage, %A95456 A92618Al base baseCu 0.04 2.12Si 0.16 0.20Mn 0.63 0.03Mg 6.80 1.56Ti 0.1 0.05Zn 0.2 0.06Fe 0.22 1.0Ni - 0.80Cr - -The values of height h exp and inner diameter d expwere determined after compression of the ringsamples. The inner diameter was measured in threelocations along the height of the rings. Finally, thevalue of the inner diameter was determined asd exp =(d top +d mid +d bot )/3, where d top , d mid and d bot arerespectively inner diameter at the top, middle andbottom along the height of the ring, accordingly.3 NUMERICAL SIMULATION OF RINGUPSETTINGA numerical simulation of the ring test was carriedout by means of the FE system QFORM-2D(QuantorForm Ltd., Russia). The aims of thesimulation were to determine the values of thefriction factor based on the obtained experimentaldata for the lubricants under study, and further toconstruct calibration curves. Levanov’s frictionmodel [6, 7] was implemented by the QFORMsimulation software.Levanov’s friction model can be considered as acombination of Coulomb’s friction law and constantfriction law. It gives almost the same results asCoulomb’s friction law for low value of contactpressure σ n . In case of high contact pressure,Levanov’s friction model and constant friction lawallow us to obtain approximately the same values offriction shear stress.Fig. 1. Scheme of calculationFigure 1 illustrates the scheme of numericalcalculation performed. Here d exp and d fem are theinner diameter of the ring sample obtainedexperimentally and by FEM, respectively.The variable parameter in the simulation was thefriction factor. The simulation was carried out forthe same values of temperature as the experimentshad been done.Stress–strain curves of alloys under study were takenfrom the handbook [8]. These curves were obtainedunder the conditions given in Table 2.Table 2. Identification of flow stress-strain curvesAlloy type T, °C ε ε& , s -1A92618 200-470 0-3 0-3A95456 150-450 0-0.8 0.01-100To perform the simulation, we assumed that thecontact friction was constant at the definedtemperature of deformation within the investigatedrange.4 RESULTS4.1 Effect of temperature, strain rate andlubricant’s compositionAccording to scheme (see Fig.1), one parameter

should be compared. That is the inner diameter ofthe deformed ring specimen.Comparing the volumes of a ring before and afterthe FE simulation showed that the loss in volumewas approximately equal to 0.5%. As a result, it isuseless to control the outer diameter before and aftereach simulation trial. That is why the inner diameteris only controlled.The determined values of the friction factor k n aregiven in Table 3. Figures 2 and 3 show therelationships between the friction factors and thetemperatures for the lubricants under study.Moreover, in Table 3 the value of friction factortaken from [5] are given. These data correspond tothe upsetting of rings made from the same Al-alloyswith the help of hydraulic press. The deformationprocess was isothermal in those tests. It means thatthe initial temperature of a specimen and tools wasequal to each other. The initial strain rate was0.14 s -1 .Table 3. Friction actors k n valuesFriction factor k nT, °C Strain rate, s -10.14 s –1 28.6 s –1 0.14 s –1 28.6 s –1A95456 A95456 A92618 A92618IO+G200 0.171 0.41 0.25 0.21350 0.153 0.287 0.236 0.202430 0.155 0.245 0.146 0.179450 0.118 0.216 0.15 0.17470 0.11 0.21 0.138 0.143SO+G200 0.245 0.45 0.19 0.34350 0.206 0.396 0.223 0.32430 0.15 0.37 0.143 0.26450 0.12 0.31 0.114 0.24470 0.143 0.285 0.133 0.235Figures 2 and 3 illustrate the effect of twoparameters on friction factor, i.e. temperature andinitial strain rate. It can be seen that the moretemperature of deformation, the less friction factorvalue is. It is valid for both investigated alloys.On the other hand, using two different presses interms of their design allowed to investigate theeffect of strain rate on friction factor value.Unfortunately, we could not perform thiscomparison within the whole range of investigatedtemperatures. The reason for that is linked to the factthat the conditions of deformation in each casesdiffered from each other. In case of using of thescrew press for deformation the isothermal conditionwas approximately observed at the temperature of200°C.Taking it into consideration it can be concluded thatat temperature of 200°C the effect of strain rate ismore significant for Al-Mg alloy. Moreover, it isindependent from the lubricant’s composition.Increasing the strain rate from 0.14 to 28.6 s -1 givesrise to increase in friction factor in over two times. Itis valid for both lubricants.Fig. 2. Deformation of A92618 alloy: solid and dash lines –regression; points - experimentFig. 3. Deformation of A95456 alloy: solid and dash lines –regression; points - experimentOn the other hand, for Al-Cu-Mg-Fe-Ni alloy theeffect of strain rate on friction factor value is not soobvious. For the lubricant based on synthetic oil(SO+G) the influence of initial strain rate on k n isalmost the same. The increase in friction factor valueis over 1.5 times. For IO+G lubricant the effect ofstrain rate is opposite to it. The difference betweenthe values of friction factor for both initial values ofstrain rate is inessential.4.2 Regression for friction factor vs temperatureThe temperature dependence of the friction factor

can be described using the following equation:2k n = Ao+ A1× To+ A2× To, (1)where A o , A 1 and A 2 = coefficients, and T о = thetemperature of the deformed material.The values of the coefficients in equation (1) aregiven in Table 4.Table 4. Coefficients of temperature dependence for k nCoefficientsStrain rate, s -10.14 s –1 28.6 s –1 0.14 s –1 28.6 s –1A95456 A95456 A92618 A92618IO+GA o , °C 0.16 0.387 0.086 0.10A 1 , 1/°C 0.00037 0.00037 0.0013 0.00086A 2 , 1/(°C) 2 -1.01×10 -6 -1.65×10 -6 -2.55×10 -6 -1.6×10 -6SO+GA o , °C 0.155 0.32 0.145 0.224A 1 , 1/°C 0.00088 0.00113 0.00085 0.00101A 2 , 1/(°C) 2 -2.05×10 -6 -2.53×10 -6 -2.03×10 -6 -2.13×10 -65 CONCLUSIONSIn summary, the following conclusions can be drawnfrom the results presented here:• The obtained temperature equations for thefriction factor can be used for FE simulation ofthe massive hot forging of A95456 and A92618alloys at strain rate of order 10 -2 and 10 1 s -1 .• The calculated calibration curves for T о = 450°Ccan be applied to estimate the friction factor of alubricant for use in the massive hot forging ofinvestigated alloys.4.3 Calibration curvesThe numerical simulation also allowed us toconstruct calibration curves. The results of FEM areshown in Figures 4 and 5, in terms of reduction ininternal diameter of ring specimens as functions ofthe reduction in their height.Fig. 4. Calibration curves - A92618 alloyThe obtained curves correspond to an initialtemperature of the deformed material of T о = 450°C.This temperature is the optimal temperature ofmassive hot forging of both aluminum alloys understudy.These plots allow to estimate the friction property ofa new lubricant which is using for deformation ofeither A92618 alloy or A95456 alloy. Temperatureof material’s specimen/tools as well as initial strainrate in the test should be the same as mentioned inSection 2.REFERENCESFig. 5. Calibration curves - A95456 alloy1. M. Kunogi, On Plastic Deformation of Hollow CylindersUnder Axial Compressive Loading. Rep. Sci. Res. Inst.(Tokyo) 2 (1954) 63-92.2. A.T. Male, M.G. Cocroft, Method for the Determinationof the Coefficient of Friction of Metals UnderConditions of Bulk Plastic Deformation. J.Instit.Metals.93 (1964-65) 38-46.3. M. Burgdorf, Investigation of Friction Values for MetalForming Processes by the Ring Compression Method.Industrie-Anzierger. 89 (1967) 7994. K.P.Rao, K.Sivaram, A review of ring-compressiontesting and applicability of the calibration curves,J.Mat.Proc.Technol. 37 (1993) 295-318.5. P.Petrov, Generalized approach to the choice of lubricantfor hot isothermal forging of aluminium alloys,Computer Methods in Materials Science 7(2) (2007)106-111.6. A.N. Levanov, V.L. Kolmogorov, S.P. Burkin, B.R.Kartak, U.V. Ashpur and U.I. Spasskiy, Contact Frictionin Metal Forging, Metallurgia, Moscow (1976).7. P.Petrov, M.Petrov, Experimental and numericalinvestigation of friction in hot isothermal deformation ofaluminium alloy AA3003, The 7th InternationalESAFORM Conference on Material Forming, Norway,Romania, Cluj-Napoca, 27-29 April, 2005, p.511-514.8. P.G. Mikliev, Mechanical properties of Light Alloys inTemperatures and Strain Rates of its Forging,Metallurgia, Moscow (1976).

Performance enhancements of die casting tools trough PVDnanocoatingsM. Rosso 1 , D. Ugues 1 , E. Torres 1 , M. Perucca 2 , P. Kapranos 31 Politecnico di Torino – Cso Duca degli Abruzzi, 24, 10129 Turin (Italy)URL: www.polito.it e-mail: mario.rosso@polito.it; daniele.ugues@polito.it;eloy.torres@polito.it2 Clean NT Lab, Environment Park - Via Livorno, 58/60, 10144 Turin (Italy)URL: http://www.cleanntlab.com/e-mail: massimo.perucca@envipark.com3 The University of Sheffield - Mappin Street, Sheffield, S1 3JD, United KingdomURL: www.shef.ac.uk/materialse-mail: p.kapranos@shef.ac.ukABSTRACT: Wear and failure of die casting dies involve a complex interaction between variousmechanisms. The most important wear and failure modes are summarized as follows: (i) the so-calledwashout damages on working die surfaces are attributed to erosion, corrosion and soldering; (ii) thermalfatigue is the most important failure mode in die casting. The sector of surface thin PVD coatings isconstantly enhancing in order to meet the increasing demand for improved performances of tooling.Advanced PVD coatings are designed to withstand severe mechanical and thermal stress conditions. ACrAlSiN nanostructured coating system was deposited on the base material, modulating the chemicalcomposition so as to increase either the chromium or the aluminium-silicon content. Then, another set ofspecimens was subjected to a cyclic immersion program in a molten aluminium bath. The washout signs weredetected and monitored in function of the increasing number of cycles.Key words: PVD ceramic coatings, aluminium die casting, washout, soldering, modulated PVD composition1 INTRODUCTIONThe process of aluminium die casting presents acomplex phenomenon of degradation, whereerosion, soldering, corrosion, thermal fatigueusually occur jointly. All of these phenomena aremajor sources of limitation to the die castingsservice life. The application of hard PVD coatingsmay result efficient against these degradationmechanisms when applied on particular parts of thedie: e.g. the inlet and the pins.The development of physical vapour deposition(PVD) has provided engineers and systemsdesigners with the ability to tailor the surfaceproperties of a range of mechanical devices to suit agrowing range of applications. Advanced PVDcoatings are designed to withstand severemechanical and thermal stress conditions.Generally, the main requirements expected byadvanced coatings are high hardness andcompression strength, high wear resistance, highmechanical and thermal fatigue resistance and lowfriction coefficient. These parameters may beattained by properly functionalising tools surfaceswith innovative targeted thin films.Furthermore, the adhesion strength at the interfacebetween coating and substrate results of majorimportance to guarantee a long endurance of the toollifetime and surface properties. In the recent years,single layer, multilayer and gradient microstructureshave been developed and now, withtesting of different coating chemistry andstoichiometry, representing the cutting edgesolutions in coating technology.The present paper reports the results of a research onthe effect of PVD films applied on a heat treated hotwork tool steel substrate. The aim of this study wasto determine the behaviour exhibited by eachcoatings in relation with the modulated chemicalcomposition. To this purpose a set of coatedspecimens with chemical composition variationsbased on the CrAlSiN system were produced

through an arc cathodic equipment. The so producedspecimens were subjected to a program of cyclicimmersions in molten aluminium alloy bath. Theassessment of damages on the cycled specimens wasperformed through optical and SEM microscopy ofboth surfaces and transverse sections, so as toanalyse the presence, if any, of soldering pits.2 EXPERIMENTAL PART2.1 Specimens fabricationA parent block was cut from an annealed AISI H11hot rolled bar, produced through vacuum meltingand remelting processes. From this test coupon a setof cubic specimens were fabricated. Thesespecimens were then vacuum heat treated accordingto the following parameters: austenitizing at 1,000°C– quenching with 5 bar nitrogen flow - firsttempering at 550°C – second and third tempering at595°C to about 45.5 HRC, which is rather typical forhot work tool steels.Ceramic coatings were deposited through thePhysical Vapor Deposition (PVD) process providedby the PL-55 prototype unit installed at the CleanNT Lab. The unit is equipped with the innovativeLateral Arc Rotating Cathodes (LARC ® ) system.Two rotating cylindrical cathodes allow enlargedtarget surface and continuous surface refreshingduring evaporation of the metallic constituentscharacterizing the ceramic coating.The CrAlSiN coating systems were deposited on theall specimens steel base material AISI H11 withdifferent modulations of the chemical compositionwere developed so as to increase either thechromium or the aluminium-silicon content. Theobtained coatings were roughly 3 μm thick. Theywere characterized by a multilayered structurecovering about 50% of the entire coating depth,followed by a complementary massive monolayer ofdefined chemical composition. Six chemicalcomposition modulations of the external layer wereprepared according to the variations of the cathodearc current and of the nitrogen flow.The coating thickness was evaluated through the ballcrater technique. A Rockwell indenter was used tostudy the adhesion properties of the deposited filmsby the indentation method.Fracture cross-section of the coated AISI H11samples was prepared to the scanning electronmicroscope (SEM) morphology investigation. X-raydiffraction (XRD) analysis was performed using aSiemens D5000 X-ray diffractometer.Monochromatic Cu-Kα radiation, produced at anacceleration voltage of 40 keV and 40 mA current,was used an initial incidence angle of 10°.Table 1. Deposition parameters used for the fabrication of thesix different CrAlSiN coating systemsCathodic arc current AlSi/Cr [A]N 2 [sccm] 110 / 70 100 / 90 80 / 100 60 / 110120 A1 D1150 B2 C2190 A3 D3The coatings hardness was determined using aMitutoyo MVK-G1 microharness tester at 100, 50,25, 10 g loads. Later the hardness of the differentcoatings was extrapolated from the evaluation ofmicrohardness at increasing weights and applying amodified form of the work-of-indentation model.HHCf− H− HSS=1+1( β β ) X0Where H c is the composite hardness, that is thehardness due to the effects of the film and thesubstrate; Hs is the substrate hardness; H f is the filmhardness; β is the relative indentation depth (RID),defined as the ratio of the maximum indenterpenetration depth to the coating thickness; β 0 and Xare systems parameters.Tribological properties of the coatings deposited onAISI H11 substrates were evaluated by means oflow frequency (5 Hz) reciprocating sliding wear test(at 5 N normal load for 500 m sliding distance) witha 10 mm diameter SAE 52100 hardened steel ball, ata temperature 25 ± 2 °C and 50 ± 5 % relativehumidity and amplitude 10mm (according to theASTM G133 standard). The wear scar was observedthrough electronic microscopy (SEM) and EDS spotanalysis was used to confirm whether the coatinghad failed or if there was simply transfer ascounterface material from the steel ball.The specimens were then subjected to cyclicimmersions in a molten aluminium alloy so as tosimulate the environmental conditions that occur atthe surface of a high pressure die casting. The(1)

aluminium alloy was AlSi 8 Cu 3 Fe, commonly usedfor die casting. The cooling bath is constituted of adiluted solution (1:80) of a commercial die lubricant.The duration of a typical cycle is of 30s with anactual immersion time of 4s for both the aluminiumalloy and cooling baths. All the specimens testedwere periodically analyzed to detect the formation ofsurface defects by SEM inspection.3 RESULTS AND DISCUSSIONThe reflected light micrograph of the ball crater(figure 1) clearly shows the multilayer structure ofthe coatings and led to a thickness measurement of2.875±0.074 μm, and clearly shows the multilayeredstructure of the coatings.Fig. 2. Adhesion Test: appearance of Rockwell indentations onspecimen A3The coatings deposited tested in reciprocatingsliding in term of friction coefficient exhibited asimilar trend, where the friction coefficient increasedgradually from 0.55 to 0.65. SEM micrographs ofthe wear scars for coatings A1, A3, B2 and C2(figure 3) shows ploughings marks along thedirection of sliding, instead for coatings D1 and D3(figure 4) demonstrate ploughings marks along thedirection of sliding plus signs of delamination areas.Fig. 1. Reflected light micrograph of a ball crater on A1The appearance of Rockwell C indentation onrepresentative coated systems is reported in figure 2.The adhesion was evaluated on the base of anobservation of the aspects of the cracks and of thepresence of coating detachments around theindentation. The adhesion evaluated through this testcan be considered good since no detachments of thecoating along the edge of the indentation could berevealed.Fig. 3. SEM micrographic wear scars specimen A1The hardness evaluations for the different coatingsas a function of increasing applied weights and, as aconsequence, of increasing indentation depth. The Acoatings presented the highest film hardness (46GPa), the D coatings the lowest (38 GPa), whereasthe B and C coatings exhibited an intermediatehardness value (43 GPa).Fig. 4. SEM micrographic wear scars specimen D1

To verify the absence of coating in the critical zones,a deepening was performed using an EDS (EnergyDispersive X-ray Spectrometer) probe (figure 4). Bythis method, the coating spoliation could beconfirmed since contents were detected in thecritical points. figure 5 and figure 6 shows one of thecritical points and details of the EDS spectrum in thetwo significant zones.after the first 5000 cycles.The cyclic immersion test in the molten aluminiumalloy clearly demonstrated that the ceramic coatingsdeposited on the steel surface may play a positiverole in terms of the reduction of the soldering effect.In the best performing systems, A1 and B2, thesoldering didn’t appear up to ca. 7500 cycles4 CONCLUSIONSFig. 3. Details of the SEM micrographic specimen D1On the base of the complete coating qualityassessments (thickness, adhesion, hardness,tribological properties), the A coatings were(systems with high content aluminum-silicon)considered the most promising candidate for thefinal applications on protection of aluminium diecasting tools. The coatings where a high aluminiumsiliconcomposition with a concurrent medium tolow nitrogen content was realized resulted to be themost effective against soldering. On the contrary, theenrichment in chromium of coating composition wasfound to be not so effective in preventing aluminiumsoldering.REFERENCESFig. 4. EDS spectrum graphs by 2 different zonesAs for the washout resistance, a first ranking of thecoated specimens performances can be drawn justafter the first step of immersion in moltenaluminium. Actually, specimens D1 and D3 (thosewith the lowest AlSi/Cr ratio) presented theformation of a thick soldered layer just after 5000cycles. On the contrary the observation of the otherspecimens did not revealed at all the presence ofsoldering points (specimens A1 and B2) or revealedvery few soldering points (specimens A3 and C2)1. Chowdhury, A., Cameron, D., Hashmi, M. “Adhesion ofcarbon nitride thin films on tool steel” Surface AndCoatings Technology, 116-119, (1999).p. 462. , D., Holler, F., Mitterer, C. “Hard coatings produced byPACVD applied to aluminium die casting” Surface AndCoatings Technology, 116-119, (1999) p.530.3. Holler, F., Ustel, F., Mitterer, C., Heim, D., Proc.“Thermal cycling and oxidation behaviour of hardcoatings in aluminium die casting” 5th Int. Conf. onTooling, Leoben, Inst. Für Metallkunde undWerkstoffprüfung, Leoben, Austria, (1999) p.357.4. Joshi, V., Srivastava, A., R. Shivpuri, R. “Investigatingtribochemical behavior of nitrided die casting surface”Proc. 6th Int. Conf. on Tooling, Karlstad, KarlstadUniversity, Karlstad, Sweden, (2002) p.809.5. Mitterer, C., Holler, F., Ustel, F., Heim, D. “Applicationof hard coatings in aluminium die casting — soldering,erosion and thermal fatigue behaviour” Surface AndCoatings Technology 125, (2000) p.233.6. J.R. Tuck, A.M. Korsunsky, D.G. Bhatc, S.J. Bull,“Indentation hardness evaluation of cathodic arc depositedthin hard coatings” Surface and Coatings Technology 139(2001).p. 63-74.7. Ugues, D., Rosso, M., Albertinazzi, M., Raimondi, F.,Silipigni, A. (2004a) “Heat treatments and innovativesurface treatments for high performing dies” Proc. 2nd Int.Conf. High Tech Die Casting, Brescia, Italy, p. 155.

An improved « plastic wave » friction model for rough contact inaxisymmetric modeling of bulk forming processesE. Vidal-Sallé 1 , S. Boutabba 2 , Y. Cui 1 , J.C. Boyer 11 Département de Génie Mécanique Conception, Institut National des Sciences Appliquées, 20 Av A. Einstein,69621 Villeurbanne Cedex, FranceURL: www.insa-lyon.fr e-mail: Emmanuelle.Vidal-Salle@insa-lyon.fr; Jean-Claude.Boyer@insa-lyon.fr2 Département de Génie Mécanique, Université de Guelma, Guelma, Algériee-mail: boutabba_s_lpg@yahoo.fr;ABSTRACT: In the case of a rough contact without friction between axisymmetric billets and punches,experimental evidences as well as very fine numerical modelling show that the workpiece material flows as alubricant layer in the vicinity of the contact surface. As no existing friction laws used for bulk formingprocesses can take into account this typical surface behaviour, an improvement of the so-called “plastic wave”friction law is proposed in order to describe more accurately the plastic flow in the workpiece-tool interface.Key words: Bulk Forming, Rough Friction, Plastic Wave Model1 INTRODUCTIONThe upsetting of billets with conical punches ofknown surface roughness and appropriate angles isproposed as a new friction test for the parameteridentification of workpiece-tool interface behaviourduring bulk forming processes. The experimentaland numerical analysis of this forming operationshowed a plastic flow of the material layer beneaththe workpiece surface that led to an “unpredictable”deformed shape of the specimen with the classicalfriction laws. So, this simple process is simulatedwith the finite element method and very fine meshesin order to predict numerically the deformed shapeof the cylinders near the workpiece-tool interfacewith the actual geometry of the tool roughness. Inorder to model the behaviour of the actual interfacewith a macroscopic law, an improvement of theplastic wave friction model is developed with tiltedboundary conditions linked to the displacements ofthe nodes sliding on the surface tool.2 THE PLASTIC WAVE FRICTION MODEL2.1 Summary of the theoryFor rough contact, Challen and Oxley [1] hadproposed a friction model using the plasticdeformation of the material workpiece in localasperities of the rigid tool surface. The model isbased upon an ideal 2D geometry of the tool surface.For the sake of simplifying the theory, the actualroughness is supposed to be equivalent to triangularwaves, height and wavelengths of which equal themean height R and the mean wavelength AR of theactual roughness profile respectively, see figure 1.R/2Ra=R/4-R/2Wawelength = ARRoughness ProfileHeight =RFig. 1. Triangular asperityThese profile parameters are defined by the standardISO 4287-1997 and can be measured withprofilometers. With this simple geometry, the flowof the workpiece material under the tool asperity isanalysed with the slip line theory. Assuming aperfectly plastic material, the value of the frictionstress τ f is given by the balance of the externaltractions, see equation (1). τ f is a function of twocharacteristic angles Φ and η of the plastic wave, σ ythe yield stress of the billets material, α the asperityα

angle, and m 0 the constant friction coefficientbetween the tool asperity and the billet surface :σ ⎡⎧ ⎛ π ⎞⎫⎤⎢ ⎜ ⎟ ( ) ⎥2 3 cosα ⎣⎩⎝ 4 ⎠⎭⎦yτf= ⎨1 + 2 + Φ − η ⎬sinα + cos α + 2Φcos m⎧⎪ sinα2 ⎪⎩1 m−10−1α + Φ = η = sin ⎨ ⎬( − )0⎫⎪⎪⎭(1)With this theoretical approach, the proposed frictionlaw for rough contact looks like a mix of theCoulomb and Tresca laws, see figure 2 where theabscissa axis is the ratio of the normal stress to thetensile yield stress of the workpiece material, and theordinate axis, the ratio of the friction stress to theshear yield stress:Normalised friction stress0.30.250.20.150.10.05Angle = 5°Angle = 4°Angle = 3°Angle = 2°Angle = 1°Angle = 0.5°0 1 2 3 4 5 6 7 8Normalised normal stressFig. 2. Plastic wave friction lawFor normal stress lower than 1.5 the tensile yieldstress of the workpiece material, the friction stressobeys to the Coulomb law with a friction coefficientsensitive to the angle α of the tool asperity. For highnormal stress, the friction stress is quite constant asproposed by the Tresca model with a constantcoefficient also sensitive to the asperity angle α. Inorder to use this interface constitutive law, threeparameters have to be identified: the yield stress ofthe workpiece material in the vicinity of the contactarea, the tool asperity angle α, and the constantfriction coefficient m 0 between the tool asperity andthe plastic wave surface. The identification of thefirst parameter needs torsion tests of the materialworkpiece as the plastic strain could reach values inthe range 100% to 500%. The asperity angle α canbe deduced from the roughness profile if the actualmesoscopic geometry of the tool surface is not toofar from the assumed triangular profile used for theslip line theory discussion. The main unknownparameter is the constant coefficient m 0 which, atpresent, is still “measured” with the so-called ringcompressiontest proposed first by Kunogi [2] forcomparison of cold forging lubricants for rollingwhere two opposite plastic flows are separated by aneutral zone under high contact pressure. Otherauthors like Petersen et al. [3], proposed alternativering geometries in order to modify this test for thelow contact pressure range. Anyway, the slidingvelocity distribution and the contact pressuredistribution at the tool-ring interface are nonuniform,see Vidal-Sallé et al. [4], and the influenceof strain hardening is not negligible, so a simple testhas to be designed for a consistent identification ofthe constant friction parameter m 0 . As a newattempt, a compression test with conical punchgeometries and low constant friction parameter m 0 isproposed in order to check the predictions of theplastic wave model on simple axisymmetric cases.2.2 Actual and theoretical roughnessFor axisymmetric geometry, the punches aremachined by turning and the plastic flow of theworkpiece is purely radial so the 2D model of theplastic wave model is pertinent. Nowadays,machining is using inserts of known circulargeometry. The turned punch surfaces were analysedwith a profilometer and the roughness profiles of thesurfaces are similar to the record of figure 3:Fig. 3. Actual profile of a turned surfaceFor this conical punch, the asperity wave length ofthe circular arc is close to 360 µm and its asperityheight close to 38 µm but one issue to address ishow to transform this circular arc geometry in atriangular profile in order to evaluate the asperityangle α. The simplest solution is a triangularasperity with the same height and wave length butthe volume of the plastic wave trapped by the toolroughness will be slightly underestimated. If themass balance must be fulfilled, the triangularasperity must have an area equal to the circular arc.This second solution leads to a height of thetriangular asperity greater than the actual one if thewave length is held constant and so to a sharperasperity angle. Numerical tests were conductedwithout local friction for five different roughnessprofiles in the range of the machined punches inorder to compare displacements, stress and plasticstrain under a conical punch modelled with its actual

circular arcs on the first hand, triangular asperitieswith equivalent height and triangular asperities withequivalent area on the second hand. From thesecomparisons, the equivalent height triangle asperityhas been found to be the better solution withdistributions of displacements and of von Misesstress very close to the distributions with the actualcircular arc profile. So, for further numerical andtheoretical investigations, the equivalent heighttriangle asperity will be used instead of the actualcircular arc.12° with different heights and wave lengthasperities, see table 1. Experiments were conductedonly with rough dies.3 EXPERIMENTAL DATA3.1 BilletsThe billets were cylindrical with an initial diameterof 60 mm and an initial length of 80 mm. Thematerial used was an aluminum alloy with a lowyield stress identified with a tensile test. The stressstraincurve was measured only up to 10% and thenextrapolated with a logarithmic function for largerplastic strain, see figure 4:Fig. 4. Stress-strain curve of the billet materialFig. 5. Experimental set upTable1. Roughness parameters of the punchesCone Angle R(µm) AR(µm) θ(°)3° 2.913 120.681 2.765° 13.842 238.970 6.6112° 38.310 358.978 12.043° 9.164 240.131 4.367° 20.985 359.060 6.673.3 Experimental resultsTwo different lubrication conditions were appliedfor the upsetting with conical punches, the first onewas dry friction and the second one “perfect sliding”with a bisulphide-molybdenum grease. Thedeformed specimens for three different roughnessesare presented on figure 6. Dry friction inducedbulging while perfect sliding led to “diabolo” likeshape. This macroscopic difference can be used toidentify friction parameter but it is not worthwhile tonotice that the displacements of the equator of thebillets depend also on other parameters than friction.3.2 Upsetting with conical punchesThe classical friction test for bulk forming processesare mainly based on compression that inducesheterogeneous strain hardening in the billets withrough punches and friction. In order to reduce theheterogeneity of the yield stress in the billets, theupsetting with conical punch is tested with particularvalues of the cone angles set equal to the asperityangle α of the triangular roughness and quasi perfectsliding. The billets are cylindrical but their ends aremachined as concave cones for matching with theupper and lower punches. Different upsetting testswere conducted for roughness ranging from 3° toFig. 6. Deformed specimens with and without frictionThe variations of the upper and lower diameters ofthe deformed billets are more representative of theworkpiece-tool interface behaviour and thedisplacements of the upper and lower billet edgeswill be the main variables used for the identificationof the constant friction parameter m 0 of the plasticwave friction model.

4 NUMERICAL MODELLING4.1 Upsetting with conical punchesThe problem symmetries allow considering only onehalf of one punch and one quarter of the billet.model and flat punches. In figure 8 both billets aregiven with the same scale. Discrepancies found canbe explained by the differences on the geometricdescription. With the rough tool, the mesh is fineenough to take into account all the asperities of theinterface. It is not the case for the left map where themeshing is coarse.Fig. 7. Plastic strain distribution in the billet for perfect slidingThe punch material is considered as elastic and thebillet material elastic-plastic with the strainhardening defined by the constitutive law given onfigure 5. In the area of contact with the punch, themesh of the billet is very fine with twelve elementsin front of each tool asperity of a 7° cone angle. Theexplicit algorithm of the Abaqus software with anappropriate mass scaling is used for solving thishuge number of equations. In spite of this choice,only small billets with 6 mm diameter and 8 mmlength were analysed in a reasonable time with adual core PC. Nevertheless, the results werequalitatively attractive as a mesoscopic effect of theroughness was observed. Some kind of Venturieffect is observed in a 0.2 mm thick layer of thebillet material with very large plastic strains (greaterthan 200%). This plastic flow feeds a flash at theedge of the billet, see figure 7. Accurateobservations of the deformed specimens presentedon figure 6 confirm this fact with a 0.7 mm largeflange on both edges of the billets.4.2 Improved plastic wave modelThe plastic flow in the vicinity of the punch surfacewas found to be sensitive to the local slope includingthe roughness profile of the tool. As it is not realisticto define the tooling geometry of any bulk formingprocess with any roughness profile, a firstimprovement would be the definition of tiltedboundary conditions coupled with the displacementsof nodes sliding on the surface tool. This solutionhas been implemented with the plastic wave frictionFig. 8. Upsetting with flat punches - Plastic strain distributionThe prediction of the displacements of the upperedge with the improved plastic wave model is veryclose to the values of the punch defined with itsactual roughness, but the deformed shape is not ingood agreement as the work hardening is found verydifferent in the two billets.5 CONCLUSIONSThe numerical analysis of an axisymmetric roughcontact with the finite element method and anexplicit integration scheme coupled to a fine meshtaking into account the roughness geometry of thetools gives results similar to the experiment. Theactual plastic flow of the billet material in the closevicinity of the tool has to be introduced in thefriction laws. Further modifications of the interfaceconstitutive equations are necessary for improvingthe numerical predictions of the billet-tool contact.ACKNOWLEDGEMENTSThe experimental part of this work has been completed underthe supervision of Pr. A. Torrance at the Department ofMechanical Engineering and Manufacturing of Trinity CollegeDublin with an Erasmus Mundus Masters grant from the E.C.REFERENCES1. J.M. Challen, and P.L.B. Oxley, An explanation of thedifferent regimes of friction and wear using asperitydeformation models, Wear 53 ,(1979) 229-2432. M. Kunogi, J. Sci. Res. Inst. .2 ,(1954), 633. S.B. Petersen, P.A.F. Martins and N. Bay, Friction inbulk metal forming, J. of Mat. Proc. Tech. 66, (1997),186-1944. E. Vidal-Sallé, L. Baillet , J.-C Boyer., Friction law andparameter identification, 2 nd ESAFORM Conference onMaterial Forming, Guimarães (1999), 603-606

Measurement of friction in a cold extrusion operation: Study bynumerical simulation of four friction testsQ. Zhang 1 , M. Arentoft 2 , S. Bruschi 3 , L. Dubar 4 , E. Felder 11 CEMEF, UMR CNRS/Ecole des Mines de Paris 7635, B.P. 207, F 06904 Sophia Antipolis Cedex, FranceURL: http://www.cemef.cma.fr/e-mail: qi.zhang@ensmp.fr; Eric.Felder@ensmp.fr2 IPU, Technical University of Denmark, Produktionstorvet, Bygning 425 DK-2800 Kgs. Lyngby, DenmarkURL: http://www.ipu.dke-mail: ma@ipu.dk3 DIMEG, University of Padova, Via Venezia, 1-35131 Padova, ItalyURL: http://www.dimeg.unipd.it/e-mail: bruschis@ing.unitn.it4 LAMIH, Université de Valenciennes Le Mont Houy F59313 Valenciennes Cedex 9, FranceURL: http://www.univ-valenciennes.fr/LAMIH/ e-mail: laurent.dubar@univ-valenciennes.frABSTRACT: The measurement of friction in the industrial metal working operations is a complexproblem because the friction test must impose at the tool/metal interface conditions similar to those in theindustrial operations. So in the European Network VIF (Virtual Intelligent Forging), a workshop (WP3) washeld to evaluate friction condition in cold forging by using numerical simulation and experimental methods. Itwas chosen an industrial cold extrusion operation, in which a low carbon steel bar, covered with phosphatelayer and soap, was drawn, cropped and then formed by extrusion. In order to know the friction condition inthis industrial extrusion operation, four kinds of friction tests, forward extrusion, double-cup extrusion,upsetting-sliding test and T-shape compression were carried out in four labs, IPU, DIMEG, LAMIH andCEMEF, respectively. In a first preliminary step we simulate the drawing and extrusion operations in order toestimate the contact conditions in extrusion along the container and the die surfaces. Then by numericalsimulation we estimate the contact conditions in the friction tests and define the parameters of the tests whichinsure the better similarity with the industrial operation. In the next step experiments will be performed inorder to compare the results of these various friction tests.Key words: Friction test, Extrusion, Cold Forging, Low Carbon Steel.1 INTRODUCTIONThe cold forging processes, including extrusion,drawing, upsetting, and heading, have been widelyused to realize the net-shape manufacture in thetransportation industry. The friction has a largeeffect on the cold forging, because it not onlychanges the forming force but also determines thesurface quality of the formed part and dies life. Theproblem is especially critical in cold forging of steelwhich induces high contact pressure and new surfacegeneration. So the billets are generally coated withzinc phosphate and soap and lubricated with oil toreduce the friction during forging [1].For reduction of the cost and time of manufacture,almost all of manufacturers and research institutesuse FE simulation to predict defects in the forgedpart and optimize the die shape. However, it is stilldifficult to choose the friction model and assign thefriction coefficient. Therefore, several methods wereinvented to evaluate the friction condition duringcold forging, such as ring compression, forwardextrusion, upsetting-sliding test, spike test, doublecup extrusion and T-shape compression. Studiesillustrated that the contact surfaces in ringcompression and spike test are shorn surfaces, whichhaven’t been coated with phosphate/soap layer, sofriction coefficient can not be adequately evaluatedby using them. Forward extrusion and double cupextrusion [2] can induce the large contact pressureand new surface generation. Meanwhile, the load offorward extrusion and cup height ratio of double cupextrusion are sensitive to friction. In upsettingslidingtest [3] friction coefficient is calculatedstarting from the tangential and normal forces of

indenter. T-shape compression, a new friction test,was developed to evaluate the friction condition byusing the formed part shape and compression load.So four suitable methods, forward extrusion, doublecupextrusion, upsetting-sliding test and T-shapecompression (see Fig. 1), can be chosen to determinefriction coefficient in cold forging. However, each ofthem has special features and they haven’t beencompared. In this paper, we investigated their abilityby FE simulation and evaluated the frictioncondition of an industrial cold extrusion process.high pressure condition, if the friction coefficient issmall.True stress (MPa)70060050040030020010000.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8True strainFig. 1 The strain/stress curve of AISI 1010 by uniaxialcompression test (drawn bar with 7.15 mm diameter)3 INDUSTRY COLD EXTRUSION PROCESS(a) Forward extrusion(b) Double-cup extrusion(c) Upsetting-sliding test (d) T-shape compressionFig. 1 Friction tests for cold forging2 MATERIAL PROPERTY AND FRICTIONLAW USED IN SIMULATIONThe material of specimen used in the simulation islow carbon AISI 1010 steel in drawn state. Therelationship between strain and stress has beenobtained by uniaxial compression test (Fig.1). Foraccurate simulation, several points on thestrain/stress curve were chosen to input into the FEcode, and we assume that the stress is constant, forstrain higher than 0.7.There are two important laws, Coulomb and(Tresca) shear stress friction law, which are appliedwidely to calculate the tangential stress betweenspecimen and tool in metal forming processes.Coulomb friction law was used in this work. It isbecause Coulomb friction law can be used in theThe whole manufacturing system includes drawing,cropping and extrusion processes. Cropping processcan induce non-uniform shorn surface of billet andsmall deformation of billet, but it is not an importantfactor on friction. Therefore only drawing andextrusion processes were considered. That is AISI1010 steel bar with 7.5 mm diameter was drawn to7.15 mm diameter and then extruded to 4.43 mm(extrusion ratio λ =2.6) in a die with a complexshape. The axisymmetrical drawing and extrusionprocess were simulated by using FORGE2005®.According to the simulation results, the maximumcontact pressure on the extrusion die can reach 2500MPa, whereas it is near 850 MPa along the container.The equivalent strain in the extruded workpiece is 2.4 FRICTION TESTS FOR COLD FORGING4.1 Forward extrusionThe operation is performed by a conical die with29.6 deg semi-angle and the extrusion ratio λequals to 1.96. The loading curve of forwardextrusion is sensitive to the friction coefficient alongthe die and along the container. So the force candecrease with decreasing the length of workpiece inthe container, whereas the die region has been filledwith the metal. However, usually the frictioncoefficients on container and die are considered asequal despite the fact that some factors alongcontainer and die are different such as material,roughness, pressure and lubricant film thickness. Fordeeply investigating the friction on the extrusion,two different friction coefficients are assigned to

contact pairs of die/workpiece and container/workpiece (See Fig 1 (a)),Fig. 3 shows the effect of friction coefficients of dieand container on the extrusion loading curves. It isseen that the extrusion load changes by assigningdifferent friction coefficient on die and container insimulation. After the extrusion experiment, thefriction coefficient of container can be defined bythe slope of the loading curve and that of die can bedefined by the maximum load.Load (KN)3530252015105µ = 0.02 µ = 0DieContainerµ = 0.07 µ = 0.0500 2 4 6 8 10 12 14 16 18 20DieContainerµ = 0.02 µ = 0.05DieStroke (mm)Fig. 3 Effect of friction coefficients of die and container on theextrusion loading curve4.2 Double-cup extrusionDouble cup extrusion test includes four parts: upperpunch, lower punch, container and specimen, asshown in Fig. 1 (b). In the test, the upper punchmoves down with the press ram while the lowerpunch and container are stationary [2]. Afterextrusion, a part with two cups was formed. Theheight of upper cup H1 is larger than that of lowercup H2, because the container has relative velocitywith respect to the upper punch, then the frictionforce from container promotes the metal flow intothe upper cup. So the ratio of the upper cup height tothe low cup height (H1/H2) was defined as cupheight ratio, which can be used to determine thefriction coefficient.For the studied configuration, the ratio of thecontainer diameter to the punch diameter is 1.66. Asin [2], we verified that the friction at the punch/billetinterface has no significant influence on H1/H2.Two kinds of material were chosen to simulateeffect of strain hardening on cup height ratio, one isAISI 1010 (see Fig.2) and another is AISI 1010without strain hardening. Simulation resultsillustrate the material strain hardening reducessignificantly the cup height ratio (see Fig.4). It isbecause the difference in strain between both metalflows increases with the cup height ratio and so thestrain hardening promotes its reduction. Therefore,Containerin order to determine the friction coefficient bydouble cup extrusion, the material property,especially the strain/stress relationship, should bewell-known and carefully assigned to FE model.H1/H2654321Without strain hardeningWith strain hardening00 1 2 3 4 5 6Stroke (mm)Fig.4 Effect of strain hardening on cup height ratio ( µ = 0.05 )4.3 Upsetting-sliding testThe basic methodology of upsetting-sliding test issimilar to the sliding indentation test. However it isspecial to simulate the real contact conditions of thedrawing and extrusion processes [3]. As shown inFig.1 (c), this test includes two main components:indenter and specimen. A cylindrical indenter playsa role of drawing or extrusion die, so the indenter ismade with the same material and surface roughnessthan the die. The specimen with coating was fixed inthe special device. Before test, the position ofindenter is adjusted to induce the penetrationdepth p of the indenter in the specimen. Then theindenter moves up with a constant tangential speedand scratches the surface of specimen. The forces inboth normal ( Fn) and tangential ( F τ) directions aremeasured with load sensors and recorded in acomputer to evaluate the friction condition betweenindenter and specimen [3].In simulation, the radius of indenter was 5 mm.Numerical simulation demonstrates that for a givenpenetration depth p , the force ratio Fτ / Fnincreaseswith the friction coefficient µ , but the contactpressure and the equivalent strain do not changesignificantly. Fig. 5 shows that the force ratio andthe equivalent strain increase with increasing thepenetration depth. The penetration depth of indenterin experiment can be decided depending on theequivalent strain of extrusion parts. Becauseequivalent strain in industrial extrusion part is 2, thepenetration depth of indenter should be chosen asp =0.5 mm (see (Fig. 5).

Tangential force / Normal force0.350.300.250.200.150.10Fτ/ Fn3.53.02.52.0Equivalent strain 1.51.00.50.050.00.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8Penetration depth (mm)Fig. 5 Effects of the penetration depth of indenter on the forceratio and the equivalent strain (µ=0.02, R=5mm)4.4 T-shape compressionT-shape compression, a new method to determinefriction coefficient in cold forging, is developed. Thedie in this test, is machined as a flat-topped die witha V-shape groove (total angle 20 deg, 7 mm depthand 1 mm entry radius) (see Fig. 1 (d)). With the flatupper punch moving down some metal of specimenwill be extruded into the groove, others will becompressed between punch and die. The sectionalshape of formed part in this test looks like ‘T’,suggesting us the name: T-shape compression test.The friction force generated between interface ofbillet and V-shape wall can directly change load andthe length H of the extrude part.The stroke and load curves with different frictioncoefficient are shown in Fig. 6 (a). Results illustratethe load increases when increasing frictioncoefficient and the large stroke can improve thesensitivity of load to friction coefficient.Load (KN)80604020µ = 0.15µ =0.10µ =0.0500 1 2 3 4 5Stroke (mm)Height of extruded part (mm)3.53.02.52.01.51.00.5Equivalent strainµ=0.05 µ=0.10µ =0.150.00 20 40 60 80Load (KN)(a) Load (b) Height of extrusion partFig. 6 Effects of friction coefficient on load and height ofextrusion partFig. 6 (b) shows effect of friction coefficient on theheight of extruded part. It is seen that the height ofextruded part increases with decrease in the frictioncoefficient. Therefore, we can determine the frictioncondition in the T-shape compression by usingcompression load the deformed part shape.5 DISCUSSION AND CONCLUSIONAccording to simulation of four kinds of frictiontests, testing conditions of them are shown in table 1.Types oftestingForwardextrusionDouble-cupextrusionUpsettingslidingtestT-shapecompression*IndustryextrusionTable 1 Testing conditions for cold forgingMaximumPressure(MPa)1800 (die)≈500 (cont.)Maximumequivalentstrain1.6 (die)0 (cont.)New surfacegeneration40 % (die)0 (cont.)800 (cont.) 1.9 (cont.) 44 % (cont.)1900 2.6 6.1 %1900 3 50 %2500 (die)850 (cont.)2 (die)0.15 (cont.)61 %0 (cont.)Concerning the main features of these tests, thefriction conditions of die and container in forwardextrusion may be different, because of differentcontact conditions. We can expect higher friction onthe die in industry extrusion because contactpressure, equivalent strain and surface generationobtain highest values in die region. For double-cupextrusion material strain hardening has a large effecton the cup height ratio and determination of frictioncoefficient. In the upsetting-sliding test, thepenetration depth of indenter can be defined by theequivalent strain of extruded part. In the T-shapecompression, the friction coefficient can bedetermined by both compression load and shape offormed part. But according to table 1 we can find itis not possible to insure similar contact conditionsfor each test and industry extrusion. So experimentswill demonstrate which contact parameters havesignificant influence on friction.ACKNOWLEDGEMENTSThe authors want to thank the VIF (VirtualIntelligent Forging - CA) project for financing theresearch of friction condition in forging.REFERENCES1. N. Bay, The state of the art in cold forging lubrication. J.Mater. Process. Technol, 46 (1994) 19-402. T. Schrader, M. Shirgaokar, T. Altan. A criticalevaluation of the double cup extrusion test for selectionof cold forging lubricants, J. Mater. Process. Technol,189 (2007) 36-443. L. Lazzarotto, L. Dubar, A. Dubois, et. al A selectionmethodology for lubricating oils in cold metal formingprocesses, Wear, 215 (1998) 1-9