A Modified Torsional Kolsky Bar for Investigating Dynamic Friction

-This paper introduces an experiment to investigate dry sliding resistance of frictional interfaces at normal pressures up to 100 MPa, slip speeds up to 10 m/s and slip distances of approximately 10 mm. This new apparatus involves a novel modification of the conventional torsional Kolsky bar apparatus, employed extensively in the past for investigating high strain rate behavior of engineering materials. The new experimental configuration represents a significant improvement over conventional tribology experiments because it uses elastic torsional waves with a superimposed static compressive force to control the interfacial traction. Moreover, the apparatus allows critical frictional parameters such as the interfacial sliding resistance, slip speeds and slip displacement to be interpreted on a microsecond time scale without the use of transducers at the frictional interface. The usefulness of the device is demonstrated by presenting results of high-speed friction on 6061-T6 AI/1018 steel and Carpenter Hampden tool steel/7075-T6 AI tribo pairs. KEY WORDS--Time-resolved friction, high-speed friction, Torsional Kolsky Bar The nature of dynamic friction forces between two bodies in contact is a complex process that is affected by a long list of factors including the interface constitution, the time scales of contact, the response of interface to normal forces, inertia and thermal effects, roughness of the contacting surfaces, history of loading and so on. Unfortunately, most experimental apparatus used to investigate dynamic friction lack the reproducibility of friction data. 1-4 This results in multibranched friction versus slip-velocity curves, which even for the same material and same experimental apparatus depend not only on the properties of the frictional interface but also on the dynamic parameters of the apparatus such as mass, stiffness and damping. In view of the scientific and technological importance of understanding dynamic friction, and given the current state of understanding, new experiments are required that can simulate the local conditions of interfacial traction, slip velocity and surface characteristics that occur in practice. These conditions need to be realized in a simple geometry for which the local traction and the slip velocity at the interface are readily measured so that more realistic models for frictional S. Rajagopalan is a Graduate Student, and V. Prakash (SEM Member) is a Professor, Department of Mechanical and Aerospace Engineering, Case Western Reserve University, Cleveland, OH 441G6. Original manuscript submitted: July 30, 1998. Final manuscript received: May 17, 1999. behavior can be developed. To accomplish these objectives, in the present study the conventional torsional Kolsky bar apparatus is modified to investigate the dynamic frictional characteristics of sliding interfaces. The modified experimental configuration allows critical frictional parameters such as the applied normal pressure, the transmitted shear stress and the interfacial slip velocity to be interpreted by the use of one-dimensional elastic wave analysis and without the use of any transducers at the frictional interface. The modified torsional Kolsky Bar friction experiments complement the dynamic friction experiments developed by Ogawa, 5 Tanimura, 6 Bowden and Persson, 7 Bowden and Freitag, 8 Prakash and Clifton, 9 Irfan and Prakash 1~ and others. Ogawa 5 modified a split Hopkinson pressure bar to study impact friction by axially impinging the input tube on a rotating output tube. Using this experimental configuration, friction experiments were conducted on a brass/brass tribo pair. In these experiments, the normal pressure was varied from 50 to 100 MPa and slip velocities of up to 5 rn/s were obtained. However the slip distance was restricted to approximately 1 mm because the duration of the experiment was approximately 200/zs. The results of these experiments indicate that when the test surfaces come in contact, almost steady-state kinetic friction is obtained within a relatively short time of 10 to 20/zs, and the measured coefficient of kinetic friction in the range of sliding velocity of up to 5 m/s is nearly constant and independent of the normal force. Higher normal pressures and higher slip velocities have been attained in the plate impact pressure-shear friction experiments conducted by Prakash and Clifton, 9 Irfan and Prakash 1~ and Prakash. 11 In these experiments, the particle velocities at the back of a target plate struck by a flyer at an oblique angle are used to infer conditions at the interface. In the experiments conducted by Irfan and Prakash, 1~ normal pressures range from 1 to 2 GPa, and slip speeds of up to 60 m/s were attained. However, because of the 3-/xs time window of these experiments, the maximum slip distance was approximately 250/zm. In their study of high-speed friction, Bowden and Freitag 8 and Bowden and Persson 7 spun a steel ball to a very high rotational speed and then grabbed it with other frictional samples or dropped it on another sample to achieve very high relative velocities (up to 800 m/s) and loads less than 0.015 MPa. In these experiments, velocity weakening of the frictional force as a function of increasing relative velocities was observed. The experiments conducted in this study involve relative sliding of dry, metallic surfaces with engineering finishes. The experiments seek to bridge the gap between the quasiExperimental Mechanics ~ 295 static friction experiments and the pressure-shear plate impact friction experiments. 1~ In these experiments, the interfacial normal pressures range from 20 to 100 MPa and slip velocities of up to 10 m/s are attained. Also, in view of the relatively large window times (approximately 1 ms) as compared to the plate impact pressure-shear friction experiments, slip distances of approximately 10 mm are obtained. Under these interfacial conditions, the constitution of the material interface is essentially stable; there is no marked penetration or normal plastic deformation of the interface, and to a large extent, from the global point of view, the frictional forces developed appear to depend on the sliding velocity of one surface relative to another. This category of dynamic friction measurement encompasses the area of interfacial friction at the tool-die workpiece interface obtained in several conventional and nonconventional material-removal and/or material-forming operations. Experimental Configuration The conventional torsional Kolsky bar apparatus consists of an incident bar and a transmitted bar supported along its length by teflon bearings. The torsional loading pulse is generated by a sudden release of stored torque. This requires a torque pulley system at the end of the incident bar and a frictional clamp positioned a short distance from the pulley end. The torque is generated by employing a hydraulic actuator to twist the pulley. The frictional clamp, designed by Hartley, Duffy and Hawley 12 for the torsional Kolsky bar, allows the desired torque to be held without slipping and releases the torque rapidly enough--when the prenotched bolt breaks-to release a sharp fronted stress pulse that travels toward the specimen sandwiched between the two bars. Two sets of strain gages are mounted on the Kolsky bar, one upstream and the other downstream of the specimen. The upstream gage monitors the incident torsional pulse and the pulse reflected from the specimen, whereas the downstream gage monitors the pulse transmitted through the specimen. The former provides a measure of the average strain rate and, by integration, the strain as a function of time. The transmitter gage provides an output proportional to the shear stress in the specimen. The schematic of the torsional Kolsky bar apparatus modified in this study to investigate dynamic friction is shown in Fig. 1. In this modified configuration, the solid incident bar appearing immediately after the hydraulic clamp in the conventional torsional Kolsky bar apparatus is replaced by a thin-walled tube. The solid portion of the bar up to the hydraulic clamp is retained to prevent crushing of the tube by the clamping mechanism. A thin-walled tubular specimen having the same cross section as the incident tube is mounted at the end of the thin-walled incident bar. The tubular specimen represents one of the materials constituting the tribe pair. We use a tubular thin-walled specimen for conducting dynamic friction experiments because (1) the thin tubular section minimizes the errors due to the averaging of the frictional stress and the interfacial slip speeds across the thickness of the specimen and (1) employing a tubular specimen of the same material and cross-sectional dimensions as the thin-walled tubular bar helps to minimize the mismatch in mechanical impedance at the incident tubespecimen interface. The transmitter bar of the conventional torsional Kolsky bar is replaced by a rigid support. Besides Solid Bar Clamp Gage Station Gage Station Bearing / r ~ Rigid N I l ~ _ , , , I ) FI ~ . . . . ~ A . T . ~ t:} ~ S u p p o r t ,1 . / / / / / / / / / / / / / Pulley | , Incident Rigi t ube ~ , ~ HsuPp~ Epoxy Tubular Specimen Fig. 1--Schematic of the modified torsional Kolsky bar employed to investigate dynamic friction providing a rigid boundary condition, the rigid support also incorporates the other half of the tribe pair. To conduct the dynamic friction experiments, the specimen on the incident bar, which has been lapped flat prior to mounting, is placed in contact with the face of the rigid support (which represents the other half of the tribe pair) by applying a static compressive force of predetermined magnitude. This axial compressive force is applied by a hydraulic actuator provided at the pulley end of the modified torsional Kolsky bar apparatus. The clamping mechanism is similar to the one used in the conventional torsional Kolsky bar, and the release of the torsional pulse is achieved by the fracture of a 606 l-T6 A1 notched pin. For each experiment, a fresh tubular specimen is epoxied onto the end of the incident bar. The mating surfaces of the tribe pair are lapped and cleaned with acetone before being compressed statically. The solid bar and thin-walled tube are 25.4 mm in diameter and fabricated of 606 I-T6 aluminum. For the experiments presented in this paper, the thin-walled tubular specimen is made from either 6061-T6 aluminum or 7075-T6 AI. The rigid support that incorporates the other half of the tribe pair is fabricated from either 1018 steel or Carpenter Hampden (CH) tool steel. The steel disk is 75 mm in diameter and approximately 25.4 mm thick. The validity of the steel disk to represent a rigid support is ensured by comparing the ratio of the mechanical impedance of the tribe pair materials, that is, the ratio of the impedance of the thin-walled tubular specimen to the impedance of the steel disk, which is approximately 1:500 for the present experiments. This condition ensures that the angular velocity of the steel disk is essentially zero and the angular velocity of the thin-walled AI specimen at the tribe pair interface is the angular slip velocity. To conduct the dynamic friction experiments, the aluminum specimen is epoxied to the end of the incident tube. Upon application of the normal force, the incident tube with the epoxied specimen slides axially in the alignment fixture and comes into contact with the face of the lapped steel disk. An important consideration in the implementation of the experiment is that while the interfacial sliding is in progress, the sliding face of the tubular specimen must remain parallel and fully in contact with the other face representing the tribe pair at all times. This is achieved by using an alignment fixture, schematically illustrated in Fig. 2, that ensures that the tubular specimen is aligned perpendicular to the other surface of the tribe pair; that is, the surfaces in 296 ~ Vol. 39, No. 4, December 1999 Steel Spacer Teflon Bearing Alignment Block ~ AI tube f i n Walled I I1.~ ~"-,,I ~ Thin walled I Incident Tube L.__i"7"~ ",~, I Tribo Pair Fig. 2--Alignment fixture contact--which are lapped flat prior to the experiment--are parallel to each other at all times. The alignment fixture has a Teflon bearing (represented by the shaded region) that allows firee rotation in either direction as well as normal motion. The body of the housing block fixture has three tapped holes to which the steel specimen (double inclined lines) is bolted. ~l]ae 75-mm steel disk, which is also one of the elements of the tribe pair, has the corresponding clearance holes for the bolts. The steel disk rests on the rigid support on hardened steel rollers to reduce hyperstiction conditions due to dry frictional effects. The face of the housing block is separated from the steel specimen by three steel spacers. The end faces of the steel spacers, the face of the housing block and the 1018 steel disk are all lapped flat. This ensures that upon assembly, the axis of the thin-walled A1 incident tube is perpendicular to the steel disk and the tribe pair faces are parallel to each other. The shear strain in the incident tube is measured by strain gages attached to the surface of the thin-walled incident tube. As the pulse travels down the incident bar, it is detected by a four-arm electric resistance strain gage (Measurement Group SK-13-120NC-10C) bridges mounted 45 deg at the strain gage station A and powered by a 30-V direct current supply. ~[qais strain gage is positioned such that there is no overlap of the incident wave and the reflected wave from the specimen end. The axial strains in the tubular incident bar are measured by means of axial strain gages attached to the incident bar at gage station B. The gage station B consists of two linear resistance strain gages that form the two opposite arms of a Wheatstone bridge, thereby helping to eliminate bending contributions. Outputs through the respective Wheatstone bridges are fed to a differential amplifier (Textronix 511A) and then onto a digital oscilloscope (TDS 420). Wave Propagation in the Modified Torsional Kolsky Bar Apparatus The wave propagation diagram is illustrated in Fig. 3. Position of the wave front versus time is detailed. The duration of the loading pulse is the time required for the pulse to travel twice the distance between the pulley and the frictional clamp. The time t --0 corresponds to the instant at which the frictional clutch is released. At this instant, the state in the solid bar is given by state 0, which corresponds to the applied torque To and zero angular velocity. The thinwalled tube to the right of the clamp has zero torque and zero angular velocity. When the clamp is released, a certain fraction of the input torque propagates as a torsional pulse to the left into the solid bar and the remaining fraction propagates to the right into the thin-walled tube. The fraction .[ Axial Force .I Torque Pulley Clamp Shear Strain Gage Tribe-Pair A L.--, \ Rigid h=1.257m 0.838m ~" L =2.439m Support Iptl Ip