j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 1 9 8 ( 2 0 0 8 ) 41–47 journal homepage: www.elsevier.com/locate/jmatprotec Characterisation of different lubricants concerning the friction coefﬁcient in forging of AA2618 B. Buchner ∗ , G. Maderthoner, B. Buchmayr University of Leoben, Department of Product Engineering, Franz-Josef-Strasse 18, A-8700 Leoben, Austria a r t i c l e i n f o a b s t r a c t Article history: In this paper a number of aluminium forging lubricants were tested with respect to their Received 27 February 2007 lubricating effect and their applicability in the production of parts with long ﬂow paths. The Received in revised form tests were conducted on a custom built testing machine that simulates the parameters of 13 June 2007 industrial production. The lubricants were assessed considering the change of the coefﬁ- Accepted 21 June 2007 cient of friction during the test and the inﬂuence on tool and workpiece surface. From the presented analysis, the most appropriate lubricant for the investigated process was found. © 2007 Elsevier B.V. All rights reserved. PACS: 06.60.Mr 06.60.Vz Keywords: Friction Forging Aluminium 1. Introduction Aluminium alloys are important construction materials in the automotive industry, in racing and aerospace due to their low speciﬁc weight, their corrosion resistance and their ability to achieve high strength with certain alloying additions. AA2618, e.g. is a heat treatable Al–Cu–Mg–Fe–Ni forging alloy developed for high temperature applications, especially for aircraft engine components (Oguocha, 1999). Due to the desired mechanical properties of aerospace components, parts made of AA2618 are typically forged. In the production of complex aluminium components by means of drop forging, besides the workpiece material, the lubricant is the most important factor to obtain a successful process. ∗ 1.1. There were two major tasks in this investigation: on the one hand, a testing method for characterising lubricants with respect to their lubricating effect and their applicability in the production of parts with long ﬂow paths had to be developed. On the other hand, experiments with a representative selection of commercial lubricants had to be performed. 1.2. State of the art In the experimental determination of friction in metal forming, direct and indirect methods are distinguished: in direct testing methods, the friction stresses are acquired by using measurement pins locally in the tool–workpiece interface, Corresponding author. Tel.: +43 3842 402 5604; fax: +43 3842 402 5602. E-mail address: [email protected] (B. Buchner). URL: http://www.metalforming.at (B. Buchner). 0924-0136/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jmatprotec.2007.06.057 Problem statement 42 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 1 9 8 ( 2 0 0 8 ) 41–47 Fig. 1 – Friction tests, where friction is determined from the change of the specimen’s shape (the specimens are marked gray): (a) ring-compression test, (b) spike test, (c) double-cup-extrusion test. whereas friction is determined via a deduced quantity such as force or deformation in indirect experiments. For the characterisation of lubricants, usually indirect methods are utilised because in indirect measurements the friction stress is averaged on the entire tool–workpiece interface and thus local inhomogeneities are compensated. Indirect testing methods, where friction is determined from specimen deformation, are very popular in industrial practice. The tests are easy to perform and represent the real process very well. However, most of the parameters cannot be controlled independently from each other, and no stationary state is present. Due to this these kind of tests are usually used in comparative investigations. The most popular friction test in this category is the ring-compression test developed by Kunogi (1954) and Male and Cockcroft (1965) (see Fig. 1(a)). This method was lately used for the evaluation of lubricants by Meiners and Hornhardt (2003) and Petrov (2007). In Fig. 1(b) the spike test utilized for lubricant characterization, e.g. by Sheljaskow (2001) is presented, and Fig. 1(c) shows the double-cup-extrusion test as depicted by Kim et al. (2004). All these experiments have in common that their parameters can be varied only in very limited ranges and the sliding distances are quite short. Indirect experiments, in which friction is determined from force or torque measurements, allow a more detailed investigation of the tribological interactions in the tool–workpiece interface. Hansen and Bay (1986) performed backward-canextrusion followed by a rotation of the container, where the torque transferred from the cup to the punch was measured (see Fig. 2(a)). Groche and Kappes (2004) utilise a sliding-upsetting test, in which at ﬁrst a cylindrical specimen is compressed with a special tool in order to obtain a homogenous surface expansion on the bottom side of the specimen and then the lower tool is moved and the sliding force is measured (Fig. 2(b)). Doege et al. (2002b) developed a model experiment to determine friction and heat transfer in backward-can-extrusion by means of a lower punch equipped with strain gauges and thermocouples (Fig. 2(c)). Lately, two different research teams (Ngaile et al., 2007; Daouben et al., 2007) evaluated lubricants by performing a sliding-upsetting test where relative motion is applied on an indenter penetrating a cylindrical billet (Fig. 2(d)). Fig. 2 – Friction tests, where friction is determined from force (or torque) measurements (the specimens are marked gray): (a) backward-can-extrusion test performed by Hansen and Bay (1986), (b) sliding-upsetting test proposed by Groche and Kappes (2004), (c) backward-can-extrusion test utilised by Doege et al. (2002b), (d) sliding-upsetting test introduced by Daouben et al. (2007) and Ngaile et al. (2007). 2. Experimental work The experiments were performed on a rotational forging tribometer (RFT) that was originally designed for ring-on-disc tests. This device had to be adapted to the actual conditions (long ﬂow paths, sliding on virgin lubricant layers) using a special toolkit. 2.1. Testing device Fig. 3 shows the RFT. The circular motion is supplied from the bottom side by means of an asynchronous motor and an angular gear box, whereas the compression force is applied by a hydraulic cylinder from the top side. The specimen (workpiece) is mounted on a rotary disc and transmits the frictional torque to the pivot-mounted ring-shaped tool. The tool is supported by a load cell via a lever arm which enables an accurate measurement of the frictional torque. The acquisition of the compression force is realised by another load cell, the rotational speed of the specimen is calculated from the speed of the asynchronous motor. Frictional speed as well as interface pressure depend on tool geometry. The specimen is heated to forging temperature using an inductive heating system whereas the tool is heated by a heating sleeve. The lubricant is sprayed onto the tool with a handgun. The testing device is controlled by a programmable logic control unit (PLC), the data acquisition is realised by a measurement ampliﬁer that is connected to a commercial personal computer. The tribometer has a maximum upsetting force of approximately 100 kN and provides a maximum torque of more then j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 1 9 8 ( 2 0 0 8 ) 41–47 43 Fig. 3 – The rotational forging tribometer testing device. Fig. 5 – Toolkit for the presented experiments. 800 Nm on the rotary disc. The rotational speed can be up to 1.7 s−1 , whereas the acceleration proceeds as shown in Fig. 4. Under this consideration, the inﬂuence of the inertia of the engine on the sliding velocity is signiﬁcant and can lead to wrong measurements, especially at short sliding distances. The sliding distance itself is not constrained by the testing device. The maximum temperature of the specimen is 1200 ◦ C, the maximum temperature of the tool is 450 ◦ C. The temperatures can be kept constant in an interval of ±2.5 ◦ C. 2.2. Toolkit and specimen geometry As the ring-on-disc test, for which the RFT was originally designed, does not meet the demands on long ﬂow paths and on the sliding at virgin lubricant layers, a new toolkit had to be developed. The sliding at virgin lubricant layers can only be realised when tool and specimen are not in contact over the entire sliding surface at the same time. This eliminates a ring-shaped geometry. Furthermore, the active surface of the tool has to be plane in order to avoid additional plastic specimen deformation leading to wrong measurements during sliding. Long sliding distances can be achieved by adjusting the distance between contact area(s) and axis of rotation. The only possibility to meet these demands, is to perform a kind of ring-on-disc test as usually used to investigate tribolog- Fig. 4 – Acceleration curves of the rotary disc for several nominal values. ical interactions at low normal pressures. However, Attanasio et al. (2007) have performed pin-on-disc experiments under forming conditions and Vergne et al. (2001) performed pin-ondisc tests to measure friction under hot rolling conditions. In both investigations, the pin represented the tool and the disc represented the workpiece. In the present analysis, the role of pin and disc was exchanged to avoid additional tangential stresses due to plastic interlocking. Furthermore, two pins were used instead of one in order to load the device symmetrically. Fig. 5 shows the toolkit used in the reported investigation, where tool and specimen were made of hot work tool steel 1.2344 (hardened to 52 HRC) and AA2618, respectively. A very important point is that the tool–workpiece contact area should remain constant during the analysis independent of the compression force. This demand was met by placing a ring around the specimen, making the contact area constant to 154 mm2 . 2.3. Execution of the tests First, the tool was brought to “operating temperature” and well lubricated. When the testing sequence was started by the PLC, the specimens were heated, and the ﬁnal temperature was kept constant for a predeﬁned time in order to allow temperature equalisation. After this, the execution of the test itself was started automatically. In order to avoid an interaction of different inﬂuences, the specimens were compressed in a ﬁrst step and thereafter the rotation began. The data acquisition was started by a mechanical trigger. After the test, tool and specimen were inspected. Three tests were executed for each parameter set. Each experiment was performed with virgin specimens, whereas the tool was changed only when a new lubricant was applied. In order to maximise the effects on a tool surface modiﬁcation, all tests were started at the same tool position. The four investigated lubricants are listed in Table 1 and the parameter matrix is shown in Table 2. It has to be stated that the specimen’s surface is enlarged during upsetting in dependence of the compression force, 44 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 1 9 8 ( 2 0 0 8 ) 41–47 Table 1 – Lubricant test matrix Lubricant type A B C Dispersion of graphite in water Dispersion of graphite in water Emulsion of a dispersion of graphite in water and mineral oil Dispersion of graphite in mineral oil D Table 2 – Parameter matrix Parameter Value Sliding velocity, v (mm/s) Normal stress, (MPa) Tool temperature, TT (◦ C) Specimen temperature, TS (◦ C) 150 210–330 250 450 e.g. different forces lead to different amounts of surface expansion. However, the inﬂuence of surface enlargement is restricted to the start of the test, thereafter virgin lubricant layers are present independently of the normal force. 2.4. Evaluation of the tests As already mentioned, the main focus of the investigations was the characterisation of different lubricants with respect to their lubricating effect and their applicability in the production of parts with long ﬂow paths. Hence, an adequate assessment was performed on the basis of the acquired measurement curves and on the basis of the inspection of the sliding surfaces after the tests. 2.4.1. Evaluation of the measurement curves During each experiment, the normal force, Fn (compression force), the tangential force, Ft (friction force), and the rotational speed, v, were measured (see Section 2.1). By means of the known contact area, A, both the normal stress, , and the friction stress, , can be calculated: = Fn , A (1) = Ft . A (2) The evaluation of the measurements was carried out in several steps (see Fig. 6): (1) All data were acquired with a ﬁxed sampling rate, i.e. the time intervals between the sampling were kept constant. However, for the comparison of measurements, a constant stroke interval is preferable. Thus, in a ﬁrst step, all the data were transferred from equal time spacing to equal stroke spacing. (2) For all three tests representing the same parameter set, average measurement curves were determined by calculating the mean -, -, and v-values for each stroke increment, i ¯ i = i1 + i2 + i3 , 3 (3) Fig. 6 – Evaluation of the measurement curves. The mean friction stress divided by with the mean normal stress gives the mean friction coefﬁcient. ¯ i = i1 + i2 + i3 , 3 (4) v̄i = vi1 + vi2 + vi3 . 3 (5) (3) The begin of the stationary region of the experiment was identiﬁed as the stroke si at which the average velocity became equal to or greater than the reference velocity (v̄i ≥ vref ). (4) For each average curve, the regression line was determined in the interval corresponding to the stationary region of the experiment. (5) From these regression lines, the values characterising the lubricants were derived. The friction coefﬁcient, , which is calculated in the centre of the regression line, is wellestablished in depicting the friction-reducing effect of lubricants = Ft , Fn (6) where Fn is the normal force and Ft is the tangential force. The characterisation of the physical stability of lubricant layers in dependence of the sliding distance is a less common task. In the scope of the present work, the following two values have been considered suitable to describe the durability of lubricant layers: the slope, ˛, of the regression line and the standard deviation, S.D., of the mean curve. 2.4.2. Surface inspection After each test, the tool and the specimen surface were assessed optically. Furthermore, after the completion of the test series, the surface roughness of all tools were measured j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 1 9 8 ( 2 0 0 8 ) 41–47 Fig. 7 – Average friction stress at one parameter set. Each curve is derived from three measurements and represents a lubricant. in order to get further information on the tribological interactions in the tool–workpiece interface. This was done by measuring roughness proﬁles in and perpendicular to the sliding direction at 20, 50 and 80 mm sliding distance and by calculating the averaged values. 3. Results and discussion Fig. 7 presents mean curves of the four lubricants listed in Table 1 at the same testing conditions. Lubricant A seems to be not affected by the surface expansion in the upset area (corresponding to the ﬁrst 14 mm of the sliding track). The sticking friction stress is high (about twice) compared to the other lubricants and it is larger than the sliding friction stress which decreases slightly with increasing distance. Regarding the lubricants B–D, the sliding friction stress exceeds the static friction stress and increases until it reaches a maximum at a distance of about 14–20 mm. However, the maximum sliding friction stress of lubricants B and C is in the range of that of lubricant A, but the sliding friction stress of lubricant D is about twice this value. At sliding distances larger than 20 mm, the friction stress reaches a steady state for lubricants B and C. Only in the case where lubricant C was tested in undiluted condition at high pressure (corresponding to the curve in Fig. 7), lubricant break-down occurred. Using lubricant D, the friction stress decreases from its maximum value and reaches a steady state at a sliding distance of approximately 50 mm. In order to show the three quantities characterising the lubricants (friction coefﬁcient, , slope of regression line, ˛, and standard deviation of the mean curve, S.D.) in a single diagram, a new quantity combining ˛ and S.D., the progression of friction coefﬁcient, pfc, was considered pfc = ˛ · |˛| 45 Fig. 8 – Results of the investigation of four lubricants. coefﬁcient is slightly decreasing with lubricant A and it is slightly increasing with lubricants B and C during the test. However, the outermost point of C indicates lubrication breakdown at the corresponding testing parameters (compare with Fig. 7). A different behaviour is shown by the oil-based lubricant D. On the one hand, the friction coefﬁcients tend to higher values than those of the lubricants on water basis, and on the other hand, the pfc is strongly negative what indicates a strong decrease of the friction coefﬁcient during the test. Fig. 9 shows the sliding tracks on the four tools. These sliding tracks were assessed visually in terms of compactness of the lubricant layer. It was found that the oil-free lubricants A and B formed compact layers that were not worn off during the test and prevented aluminium pick-up. However, after the test, parts of the lubricant layers ﬂaked off. In contrast, the oil-containing lubricants showed another behaviour. The water-oil-based lubricant C formed a very homogeneous layer, but it was worn off at high pressures and pick-up occurred in small areas. The oil-based lubricant D was completely abraded 2 ˛2 + (10 · S.D.) . (7) The factor 10 was introduced to make ˛ and S.D. of the same order of magnitude (to cause a balanced weighting). Fig. 8 shows the result of the investigation on the four lubricants. Each point in the diagram corresponds to the result of a single testing condition. The lubricants A–C show a stable behaviour (a smooth curve progression), whereas the friction Fig. 9 – Sliding tracks for different lubricants (sliding was performed clockwise). 46 j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 1 9 8 ( 2 0 0 8 ) 41–47 Table 3 – Tool roughness values Rz (m) Lubricant Sliding direction A B C D 7.58 3.81 3.72 5.03 Perpendicular 14.19 7.29 6.54 9.82 and heavy pick-up was present in the beginning of the sliding track (near the upset area). When sliding progressed, further pick-up occurred in small areas. Comparing the roughness values of the tools (see Table 3) the following is found: lubricant A gives the highest tool roughness, the roughness when applying B or C is approximately the half, and the roughness when applying lubricant D is inbetween the roughness values of A and B, respectively. The intention of this investigation was the assessment of lubricants with respect to their lubricating effect and their applicability in the production of parts with long ﬂow paths. This means in terms of the analysed quantities that the values of the friction coefﬁcient, , the slope of regression line, ˛, and the standard deviation of the mean curve, S.D., should be small. The claim for a small friction coefﬁcient in the die cavity is evident in order to achieve die ﬁlling, a small slope of regression line indicates steady state sliding conditions (no lubrication breakdown, etc.) and a small standard deviation ensures process stability. Due to these considerations, lubricant A or B would be the right choice under the investigated conditions. However, the surface roughness is much lower with lubricant B what leads to lower friction coefﬁcients and a smoother workpiece surface. The poor result of the oil-based lubricant D can be explained by regarding Fig. 9. It can be clearly seen that the tool temperature was too low to evaporate the carrying agent. As a result, the lubricant was squeezed out of the tool–workpiece interface in the upsetting step, and in the beginning of the rotation, pick-up occurred. In further sliding, hydrodynamic lubrication was present and the friction coefﬁcient decreased rapidly. However, the lubricating effect was not strong enough to prevent further pick-up completely. It is assumed, that the oil-containing lubricants C and D would give better results at higher tool temperatures. 4. Conclusion A new approach for the characterisation of lubricants under aluminium forging conditions has been presented. Compared with other tests, the demands on the lubricants are obviously less severe in the presented setup. At the beginning of the rotation, new lubricant gets into the tool–workpiece interface, and after a sliding distance of approximately 14 mm, tool and specimen are separated by a compact lubrication layer even under high initial surface expansion. Furthermore, each point of the lubricant layer is loaded just for a short time. In contrast to other experiments (see e.g. Hansen and Bay, 1986), the new test does not investigate the expansibility or the chemical stability of lubricants, but it analyses the abra- sion resistance of lubricant layers. As the lubricant is applied to the tool in hot die forging and the workpiece is in sliding contact with the tool surface (usually superimposed by surface expansion), this behaviour should be of interest in tribological analyses. Concerning the main features, the presented setup is similar to the sliding-upsetting test (Groche and Kappes, 2004), however, the new setup works without an adjustability of surface expansion, because its inﬂuence is only present in the beginning of the test. In contrast to the sliding-upsetting test, in which the sliding distance is short compared to the upset area, the new method allows the development of a – also under tribological aspects – stationary region. From the test results a clear formation of lubricant groups was decided (see Fig. 8), which allowed to determine the applicability of the lubricants for the chosen process parameters and materials. However, an additional investigation of these inﬂuences is advisable because of the interactions of sliding, surface enlargement and thermo-chemical exposure in the real process. Acknowledgements The authors want to thank the federal province of Styria (“Zukunftsfonds Steiermark”, Project 19) for ﬁnancing the project and Fuchs Schmiermittel GmbH and Acheson Industries for the provision of the lubricants. references Attanasio, A., Fiorentino, A., Ceretti, E., Giardini, C., 2007. Experimental device to study surface contacts in forming processes. In: Proceedings of the 10th ESAFORM Conference on Material Forming, pp. 275–278. Daouben, E., Dubar, L., Dubar, M., Deltombe, R., Dubois, A., Truong-Dinh, N., Lazzarotto, L., 2007. Friction and wear in hot forging of steels. In: Proceedings of the 10th ESAFORM Conference on Material Forming, pp. 505–508. Doege, E., Alasti, M., Schmidt-Jürgensen, 2002b. An innovative procedure for the numerical identiﬁcation of accurate friction and heat transfer laws for precision forging processes. In: Proceedings of the Ninth ISPE International Conference on Concurrent Engineering: Research and Application—Advances in Concurrent Engineering, pp. 199–207. Groche, P., Kappes, B., 2004. Tribologie der Massivumformung—Modellprüfstände der Tribologie. In: Bartz, W.J. (Ed.), Tribologie und Schmierung bei der Massivumformung, vol. 13. Expert Verlag, Renningen-Malmsheim, Germany, pp. 1–14. Hansen, B.G., Bay, N., 1986. Two new methods for testing lubricants for cold forging. J. Mech. Work. Technol. 13, 189–204. Kim, H., Padwad, S., Altan, T., 2004. Evaluation of new lubricants for cold forging without zinc phosphate coating. In: Proceedings of the 37th Plenary Meeting of the International Cold Forging Group. Kunogi, M., 1954. On plastic deformation of hollow cylinder under axial compression. Rep. Scient. Res. Inst. 30, 63–92. Male, A.T., Cockcroft, M.G., 1965. A method for determination of the coefﬁcient of friction of metals under conditions of bulk plastic deformation. J. Inst. Metals 93, 38–46. Meiners, F., Hornhardt, C., 2003. Die forging of aluminium—innovative solutions for industrial forging j o u r n a l o f m a t e r i a l s p r o c e s s i n g t e c h n o l o g y 1 9 8 ( 2 0 0 8 ) 41–47 operations. Part II. Establishment of suitable lubricants and tool parameters. Aluminium 79, 142–149. Ngaile, G., Saiki, H., Ruan, L., Marumo, Y., 2007. A tribo-testing method for high performance cold forging lubricants. Wear 262, 684–692. Oguocha, I.N.A., 1999. Characterization of aluminum alloy 2618 and its composites containing alumina particles, PhD thesis, University of Saskatchewan. 47 Petrov, P.A., 2007. Generalized approach to the choice of lubricant for hot isothermal forging of aluminium alloys. Comput. Meth. Mater. Sci. 7, 106–111. Sheljaskow, S., 2001. Tool lubricating systems in warm forging. J. Mater. Process. Technol. 113, 16–21. Vergne, C., Boher, C., Levaillant, C., Gras, R., 2001. Analysis of the friction and wear behaviour of hot work tool scale: application to the hot rolling process. Wear 250, 322–333.