Lowering Friction in Timing Chain Drive Systems by Tuning Tensioner Materials.
One of the major challenges for the automotive industry is meeting emission legislative requirements. This is also driven by customer demands on reduced fuel consuming cars. The main concept to reduce fuel consumption is decreasing the overall weight of the car. The overall car weight is continuously increasing over the last years however. This results from increased safety and comfort demands.
The use of lighter materials is preventing an even stronger weight increase. Another route is the development of more efficient engines for reduced fuel consumption. One way to improve engine efficiency is the reduction of friction. Reducing mass of components can decrease the friction due to lower dynamical loads. This is mainly achieved by using lighter materials or better designs for moving components.
This paper focusses on reducing sliding friction within the valve train drive system by tuning the intrinsic tribological properties of plastics used for chain tensioner arms and guides. Within valve train drive systems, timing chains are used in roughly half the engines that come to the market in Europe. To maintain the chain tension throughout the engine's life, a tensioner arm is used at the slack side of the chain. In addition, guides and tensioner arms are used to prevent lateral resonance in the chain.
A breakdown of the friction losses in a typical timing chain system of an 13 or 14 engine is shown in Figure 1. Nearly 45% of the losses are attributed to chain to tensioner arm and guide contacts.
FRICTION IN A CHAIN-ON-GUIDE CONTACT
An example of a timing chain system is shown in Figure 2. This particular system incorporates one tensioner arm and three guides. A chain tensioner arm is used at the slack side of the chain to maintain the tension throughout the engine's life. Chain guides are installed between sprockets, to prevent lateral resonance in the chain parts.
The tribology of these contacts is not only determined by the chain, tensioner arm, and guide surfaces, but also by the engine oil. Engine oil is sprayed onto the chain to reduce friction due to chain articulation and migrates to the chain to tensioner arm and guide contacts as well.
The friction in this lubricated contact depends on conditions such as chain tension, temperature, and sliding velocity. Friction under lubricated conditions is commonly represented by a Stribeck curve , which is schematically shown in Figure 3. On the vertical axis, the (dynamic) coefficient of friction ([mu]) is plotted, which is defined as the friction force at the contact divided by the normal force. On the horizontal axis the Hersey number is plotted. This number is defined as the product of sliding velocity (v) and oil viscosity ([eta]) divided by normal load (p).
When the sliding velocity is low, the pressure in the oil film is unable to support the normal load induced by the chain force and asperities on both surfaces are in direct contact with each other. Friction characteristics in this regime are quite similar to those of friction under dry conditions, although friction levels are generally much lower due to the presence of a very thin layer of oil. This domain is termed the boundary lubrication regime. When the sliding velocity increases, the pressure in the film increases and the film is able to support a larger portion of the normal load. The distance between the surfaces increases and asperities are gradually separated from each other. As a consequence, friction drops. The overall measured friction is now a combination of asperity-to-asperity friction and lubricant film friction. The velocity at which friction starts to drop depends on local (surface) parameters such as surface topology and roughness, but also global geometrical parameters such as radius of curvature of the guide and the chain links and modulus of these components. This velocity domain is termed the mixed lubrication regime.
Further increasing the sliding velocity leads to a further increase of oil film pressure enabling it to support the full normal load of the chain and separating the surfaces completely. This condition is reached near the minimum in the Stribeck curve. At even higher velocities, the friction increases again due to the viscous drag in the oil. This velocity domain is termed the (elasto-)hydrodynamic regime. In this regime, the main friction determining parameters are the oil viscosity and the moduli of the tensioner arm, the guide, and the chain.
Friction in a real timing chain drive system is far more complex. The conditions along the chain tensioner arm and guide may vary (see Figure 4) Consider for example location 1 in this Figure. The radius of curvature of the arm is relatively large and the normal load relatively low. At sufficiently high sliding velocity, friction may be in the mixed lubrication regime at this location as indicated by the red dot in the Stribeck curve. At another point (for example location 2), the normal loading conditions may be much more severe due to the smaller radius of curvature of the tensioner arm or due to dynamics in the chain. Consequently, friction at this location may be in the boundary lubrication regime. The total friction measured over the full chain drive system is a weighed sum of the many individual contributions at all locations and this leads to a very complex relation to parameters such as chain tension, sliding velocity and temperature.
The friction losses are usually determined from the measured drive torque at one of the sprockets. These losses thus also include those due to chain articulation, chain to sprocket contact, in bearings, and at cam-follower interfaces. The interpretation of the friction losses is even further complicated by edge effects at the chain links (stress concentrations, see Figure 10). (small) defects of the chain links due to stamping, and variations in the axial chain force and sliding velocity.
The ball-on-pyramid test set-up (see Figure 5) is used here in support of deepening the understanding of the mechanisms of lubricated friction and for identifying the most promising routes for lowering friction by adaptation of the chain tensioner arm and guide materials.
A sketch of the testing principle is shown in Figure 6. For a single test, 3 identical samples with dimensions 4 x 6 x 15m[m.sup.3] are placed under a 45[degrees] orientation angle as shown in the Figure. A chromium steel ball (12.7mm diameter) is placed in the center and is supported by the three samples. A normal load is applied and the ball is rotated around its vertical axis. The torque required to rotate the ball is logged and the coefficient of friction is derived from this torque. The contacts are submerged in oil, and the temperature of the oil is controlled via a heating stage.
The test allows for a fairly straightforward analysis and the test conditions are well-defined and well-controlled. However, the correlation between this test and a real chain tensioner arm contact is quite complex and not fully understood, so care should be taken to extrapolate results of this test to the real application. Nevertheless, it captures some of the essentials in a chain to tensioner arm or guide contact, i.e. converging wedge, steel and plastic surface contact in the presence of a lubricating oil.
Performing a test requires only a small amount of sample material in a simple geometry which makes it easy to screen a number of materials. The samples can be taken from molded parts or standard tensile test bars.
An important parameter in friction is the normal pressure. Using Hertzian contact mechanics, the maximum pressure at the three contact points is estimated as a function of normal load on the ball (see Figure 7) for PA66 and PA46 at 90[degrees]C. Note that the maximum pressure is dependent on the modulus of the tested material and consequently varies with material type and temperature (see Table 1 for modulus values used in the calculation). The standard normal load used in the tests described in this paper is IN. For this value, the maximum contact pressure is nearly 12 MPa for PA66 and nearly 16 MPa for PA46.
The calculated diameter of the contact spot is shown in Figure 8. At 1N normal load the estimated contact diameter is roughly 230[micro]m to 270[micro]m (depending on the type of plastic material).
The test set-up does not allow for real-time measurement of the contact diameter under the normal load. Figure 9 shows the wear mark on a Stanyl HGR2 sample (PA46 with small amounts of friction modifiers) after a friction test (normal load 1N, 90[degrees]C oil temperature, Castrol Edge 5W30 engine oil). The diameter of the wear mark is estimated at 250[micro]m to 300[micro]m which is in good agreement with the calculated values.
COMPARISON OF CHAIN-ON-GUIDE AND BALL-ON-PYRAMID
Although the ball-on-pyramid test is of different geometry (point contact) than an actual chain on guide set-up (line contact), it captures a many of the relevant characteristics of friction under oil lubrication. The main conditions relevant for friction of the two setups are compared in Table 2.
Generally, the conditions of both set-ups are quite similar except for the sliding velocity. The maximum attainable sliding velocities for the ball-on-pyramid set-up is on the low boundary of those encountered in an actual chain-on-guide contact.
The calculated contact pressure in a chain-on-guide contact given in Table 2 is merely a first order magnitude estimate. Chain tension varies during operation (dynamics) leading to fluctuations in contact pressure. Furthermore, the estimated contact pressure is based on a simple 2D configuration (Hertzian contact mechanics ), whereas in reality the (3D) stress distribution is far more complex with peaks at the edges of the chain links (see for example Figure 10 and Figure 11). Estimating the true contact pressure is even further complicated by the fact that the chain link edges are not perfectly straight and have fairly poorly defined radii of curvatures.
For these reasons, some liberty should be taken into account when correlating ball-on-pyramid results to actual timing chain friction tests.
The average drive torques measured in a motorized engine test on a Ford 2.4L Duratorq TDCi are shown in Figure 12. In this test, the engine oil was heated to 90[degrees]C. For the situation where all guide and tensioner arm faces are in PA66, the measured drive torque is larger in the majority of the engine speed domain than for the situation with guide and tensioner arm faces in PA46 (TW341).
Performing a comparison between these two materials for the ball-on-pyramid set-up, gives the results shown in Figure 13. Also in this test, the coefficient of friction of PA46 is lower than PA66 over large velocity ranges. Taking into account the complex relation between the ball-on-pyramid test and the motorized engine test discussed before, the overall agreement is encouraging.
UNDERLYING PHYSICS DETERMINING THE ENGINE OIL LUBRICATED STEEL TO PLASTIC FRICTION
This section describes a few of the studies carried out for identifying the sensitivity of a number of parameters on the measured friction in the ball-on-pyramid test.
The effect of viscosity on the measured coefficient of friction is shown in Figure 14. The lubricants used for these particular experiments were general purpose silicone viscosity standards  and the steel ball was sliding over a PBT plastic surface at 30[degrees]C lubricant temperature.
With increasing viscosity, it becomes easier to separate the two surfaces by a lubricating film due to the higher film pressure. The transitions from boundary lubrication to mixed lubrication and to hydrodynamic lubrication shifts to lower sliding velocities with increasing oil viscosity. In the hydrodynamic regime, the coefficient of friction increases more steeply for higher viscosity liquids (higher viscous shearing).
Normalizing the velocity axis with the fluid viscosity results in a 'mastercurve' (see Figure 15). To a reasonable approximation, the velocity at the transition from mixed to hydrodynamic lubrication scales with the lubricant viscosity. In the hydrodynamic regime, friction is also proportional to the lubricant viscosity.
For obtaining low values of the coefficient of friction, it is important to avoid direct contact between the two surfaces. The influence of the roughness of the plastic surface is shown in Figure 16. Increasing the surface roughness shifts the Stribeck curve in the direction of higher sliding velocity. In the boundary lubrication regime at the lowest sliding velocities, the coefficient of friction is lower for the sample with higher surface roughness. This is believed to be caused by a reduction in real contact area for a rough surface, which lowers friction .
Images of the surfaces of the smooth and the rough PA46 samples are shown in Figure 17 and Figure 18. The [R.sub.A] value of the rough surface is approximately 40 times higher than that of the smooth surface.
The roughness of the steel counter surface plays a role as well. An example of a representative portion of the surface is shown in Figure 19. The surface of the ball used in this test has been compared to that of a number of timing chain surfaces. The actual chain surfaces generally have factor 2 to 4 higher roughness values ([R.sub.A], see Table 21 and the surface topology is generally less homogeneous due to stamping. The higher surface roughness is a possible reason for the minimum in the Stribeck curve to shift to higher sliding velocity for real timing systems.
In order to avoid direct contact between the two surfaces, the film thickness required to carry the normal load must be larger than the typical height of the asperity peaks on the surfaces. The calculated film thickness [2, 7] for the experimental conditions of Figure 16 is shown in Figure 20.
The calculations show that around 1m/s sliding velocity, the required film thickness is on the order of a few 0.1 [micro]m, so on the same order of magnitude as the typical combined surface roughness for the smooth sample and the steel ball. This is in good agreement with the minimum in the measured Stribeck curve for this material. On the other hand, for the rough sample, a minimum film thickness of at least a few [micro]m would be needed which is reached at sliding velocities > 10 m/s. This is also in agreement with the trend observed in the measurements.
A large body of literature is available on determining the optimal surface structure for lowest coefficient of friction. Overviews are readily available in literature (see for example ). For the limited number of surface topologies studied here, no clear trends were identified so far between coefficient of friction and topology apart from the roughness level (peak height) described before.
Modern engine oils are complex fluids with well-engineered additive packages that interact with surfaces for example to lower wear and friction, modify viscosity, and inhibit oxidation. The influence of the oil on the measured coefficient of friction is shown in Figure 21.
When comparing the fresh engine oil to the mineral oil, there is a large difference in coefficient of friction in particular in the boundary and mixed lubrication regime where there is contact between asperities on the two surfaces. The difference is attributed to the additive package in the engine oil interacting with the metal and/or the plastic surface. The exact mechanisms are not fully understood yet. The minimum in the curve for the mineral oil appears to occur at higher speed which is probably attributed to the somewhat lower viscosity of this oil at 90[degrees]C (5W30: 13.5mPas, mineral oil 8.3mPas).
The coefficient of friction for the used engine oil is significantly higher as that of the fresh engine oil. This could be explained from theories  stating that soot particles adsorb the additives and thereby deteriorating the effectiveness of the lubricant.
The friction response of neat PA46 (Stanyl TW300) is shown as a function of temperature in Figure 22 for the Castrol Edge 5W30 engine oil and in Figure 23 for the mineral oil.
For the base mineral oil without additives, the coefficient of friction in the boundary lubrication regime steeply increases with temperature. One possible explanation is the lower modulus of the plastic at higher temperature. Assuming that friction is determined by real contact area between asperities , a lower modulus of the plastic increases the true contact area and thereby the coefficient of friction. It could also be due to the lower viscosity of the oil. Although friction in the boundary regime is generally considered to be independent of the bulk viscosity, temperature dependent reological behavior of thin lubricant films at the asperities could be responsible for this increase in friction.
The additive package in the engine oil seems to interact with the surfaces and lowers the coefficient of friction in the boundary and mixed lubrication regimes. Furthermore, this is (partly) a temperature activated process because the friction levels at 20[degrees]C are quite comparable for the two oils.
In the high velocity regime, modulus of the plastic and oil viscosity are two important parameters. Increasing the temperature lowers the modulus of the plastic and the viscosity of the oil. Lowering the viscosity of the oil would shift the minimum of the Stribeck curve to higher velocities whereas lowering the modulus of the plastic would shift it in opposite direction. A calculation shows that the required film thickness decreases with increasing temperature (Figure 24), suggesting that the minimum in the Stribeck curve shifts to higher velocities with increasing temperature. This is in line with the measurements. Apparently, the strong reduction of oil viscosity cannot be compensated by the relatively moderate reduction in modulus.
Figure 25 and Figure 26 shows the data of Figure 22 and Figure 23 with the horizontal axis corrected for the oil viscosity. The curves at different temperatures do not fall completely onto a single mastercurve. The transition from mixed to boundary lubrication reasonably scales with (temperature dependent) lubricant viscosity. The differences in the boundary lubrication regime are most likely caused by factors such as the temperature dependence of the modulus of FA46 and the additive package in the engine oil.
Lowering Friction under Engine Oil Lubricated Conditions
The ball-on-pyramid set-up was used for a screening of a number of modified FA46 materials engineered for lowering friction in particular in the boundary lubrication regime. In this regime, friction is determined by direct asperity contacts. The focus was on performance assessment of several friction lowering additives and their possible synergistic effects. This resulted in Stanyl HGR2, a PA46 base material with a combination of friction modifying additives including PTFE.
The ball-on-pyramid results for this new material and the standard PA66 and FA46 is shown in Figure 27. A significant reduction in friction has been achieved in the boundary lubrication regime.
VALIDATION OF FRICTION PERFORMANCE CLAIMS IN MOTORIZED ENGINES
Stanyl HGR2 shows a clear lowering of friction levels compared to PA66 and PA46 in particular in the boundary lubrication regime of the ball-on-pyramid test. Since the correlation with actual chain tensioner arms in engines is semi-quantitative, proof of performance enhancement should be given by performing tests on a set-ups that are more representative for the actual application.
Figure 28 and Figure 29 show results of motorized engine tests performed by BorgWarner Inc. on a Ford gasoline engine. All tests were performed using a 5W-20 engine oil which was heated to 93 [degrees]C [+ or -] 2[degrees]C. Figure 28 shows the benefit of Stanyl HGR2 over PA66 and PA46 (Stanyl TW341). The sliding speed dependence of the crankshaft torque is quite different from that in the ball-on-pyramid test (see Figure 27), illustrating again the complexity of directly translating results from one test to another.
Figure 29 shows the measured crankshaft torque of tests performed on the same motorized engine for FA46 and Stanyl HGR2. This Page 9 of 12 graph also shows the further improvement due to geometrical and tuning optimization.
CONCLUSIONS AND OUTLOOK
In finding options for lowering friction losses, the timing chain system of an engine was considered. In particular, the chain to guide and tensioner arm contacts was investigated. Focus was on the properties of the plastic chain guide and tensioner arm materials and their role in lowering parasitic friction losses.
A lab-scale test (ball-on-pyramid) was used for fundamental studies into a lubricated steel to plastic sliding contact. Although the detailed relation between this lab-scale test and the actual chain drive application is quite complex, comparisons to motorized engine tests showed qualitative agreement.
A fundamental study into the physics of lubricated friction was carried out and the effect of a number of parameters was demonstrated. The influence of the additive packages in engine oil and the age of the engine oil was shown to invoke major changes in friction.
The gained insights were used to develop a new low friction PA46 based grade. This grade (HGR2) showed marked improvements in friction both compared to standard PA46 and PA66 especially in the boundary lubrication and mixed lubrication regimes. This was confirmed in the lab-scale ball-on-pyramid test as well as in motorized engine tests performed by BorgWarner Inc. Ford will include Stanyl HGR2 in the options to improve the fuel economy of its future engine platforms.
Future work will include a more detailed exploration of the interactions between the oil and plastic surfaces in order to find additional means for lowering friction by adaptations to the plastic chain tensioner arm material.
[1.] O'Shea, F. and Sisson, J., Chain cam drive efficiency optimization and comparison to belt drives, BorgWarner Morse TEC Ithaca, Engine Expo. 2012.
[2.] Bhushan, B., Introduction to Tribology (New York: John Wiley & Sons. 2002).
[4.] Johnson, K.L., Contact Mechanics, (Cambridge: Cambridge University Press, 1985), 84
[5.] Van Ruiten, J., Proost, R. and Maile, K., "Camshaft drive optimization", Engine Technology International, January 2011, 66-68.
[7.] Dowson, D. and Higginson, G.R., Elastohydrodynamic Lubrication, (Oxford: Pergamon, 1966).
[8.] Gropper, D., Wang, L. and Harvey, T.J., "Hydrodynamic lubrication of textured surfaces: A review of modeling techniques and key findings", Tribology International, 94, 509-529, 2016.
[9.] Acros Organics mineral oil 12400020, http://www.acros.com
[10.] Antusch, S., Dienwiebel, M., Nold, E., Albers, P. et al., "On the tribochemical action of engine soot", Wear, 269, 1-12, 2010.
[11.] Bowden, F.P. and Tabor, D., The friction and lubrication of solids, (Oxford: Oxford University Press, 1954).
Jippe van Ruiten.+31 6 514 29 614.
DSM Engineering Plastics, PO Box 1077, 6160 BB Geleen, The Netherlands
The authors would like to acknowledge Fenton O'Shea of BorgWarner Inc. (
firstname.lastname@example.org) for providing the results of the motorized engine tests and his helpful discussions in preparing the manuscript.
DAM - dry as molded
[eta] - dynamic viscosity [Pas]
I3 - inline 3 cylinder engine
I4 - inline 4 cylinder engine
[mu] - coefficient of friction (friction force divided by normal force) [-]
PA46 - polyamide 46
PA66 - polyamide 66
PBT - polybutylene terephthalate
v - sliding velocity [m/s]
APPENDIX A: HERTZIAN CONTACT MECHANICS
Hertzian contact mechanics theory  is applied here to estimate the maximum pressure and radius of the contact between the steel ball and a flat plastic sample (see Figure 30) In this theory, the contacting bodies are assumed to exhibit linear elastic behavior and the indentation depth is small compared to the radius of the ball.
The pressure distribution in the contact of a ball on a flat elastic sample has a parabolic shape. The pressure reaches its maximum p0 in the center of the contact and it decreases to zero at the edge of the contact. The maximum pressure p0 is calculated from:
"[mathematical expression not reproducible]" (1)
where R is the radius of the ball, [F.sub.nc] is the normal force on the ball, and E' is the equivalent modulus of the contacting materials, defined as:
"[mathematical expression not reproducible]" (2)
where [v.sub.PL] and [E.sub.PL] are the Poisson's ratio and Young's modulus of the plastic sample and [v.sub.ST] and [E.sub.ST] those of the steel ball. Because the modulus of the steel is much larger than that of the plastic, the equivalent modulus can be approximated by:
"[mathematical expression not reproducible]" (3)
The contact radius a is calculated from:
"[mathematical expression not reproducible]" (4)
APPENDIX B: ELASTOHYDRODYNAMIC CALCULATION
In order to calculate the thickness of the oil layer between a steel ball and a flat plastic surface, the deformation of the plastic cannot be neglected and an elasto-hydrodynamic approximation has to be made. The minimum film layer thickness [h.sub.min] is calculated from the following approximating equation :
"[mathematical expression not reproducible]" (5)
where R is the radius of the ball and U is the normalized velocity defined as:
"[mathematical expression not reproducible]" (6)
where [[eta].sub.0] is the lubricant viscosity at environment pressure, u is the sliding velocity of the ball, and E' is the equivalent modulus as defined in Eq. (2).
W denotes the normalized load:
"[mathematical expression not reproducible]" (7)
where [F.sub.nc] is the normal force on the steel ball (see Figure 30).
DSM Ahead B.V.
Jippe Van Ruiten
DSM Engineering Plastics
Thijs Besseling and Robbert van Sluijs
DSM Ahead B.V.
Maik Broda and Brian Pearce
Ford Motor Company
Fenton I. O'Shea
BorgWarner Morse TEC
Table 1. Modulus (DAM) of PA66 and PA46 at several temperatures. The modulus of HGR2 is nearly the same as that of PA46. Temperature PA 66 PA46 20[degrees]C 3.4 GPa 3.3 GPa 90[degrees]C 0.7 GPa 1.0 GPa 130[degrees]C 0.5 GPa 0.7 GPa Table 2. Comparison of chain-on-guide set-up and ball-on pyramid set-up. Chain-on-guide Contact Line contact Radius curved surface 100 - 1000mm Temperature 80[degrees]C-140[degrees]C Contact pressure lMPa-20MPa Sliding velocity 1 m/s - 15m/s Material Chromium steel Modulus at 25 [degrees]C ~200 GPa Surface roughness [R.sub.A]~0.2-0.4[micro]m Ball-on-pyramid Contact Point contact Radius curved surface ~6.3 mm Temperature 25[degrees]C-150[degrees]C Contact pressure 10MPa-20MPa Sliding velocity [10.sup.-4]m/s-1.5m/s Material Chromium steel Modulus at 25 [degrees]C ~200 GPa Surface roughness [R.sub.A]~0.1[micro]m Table 3. Parameter settings for calculating the minimum oil film thickness as a function of temperature. Oil viscosity Modulus Temperature (Castrol Edge 5W30) (PA46) 20[degrees]C 150 mPas 3.3 GPa 90[degrees]C 13 mPas 1.0 GPa 130[degrees]C 6 mPas 0.7 GPa
|Printer friendly Cite/link Email Feedback|
|Author:||Meuwissen, Marcel; Ruiten, Jippe Van; Besseling, Thijs; van Sluijs, Robbert; Broda, Maik; Pearce, Br|
|Publication:||SAE International Journal of Fuels and Lubricants|
|Article Type:||Technical report|
|Date:||Apr 1, 2017|
|Previous Article:||The Relative Importance of Fuel Oxidation Chemistry and Physical Properties to Spray Ignition.|
|Next Article:||Numerical Simulation of Gasoline and n-Butanol Combustion in an Optically Accessible Research Engine.|