This section of the literature review has been dedicated to discussing the contribution of the CVT lubricant to its performance, since this is a critical factor affecting the temperature dependence of the loss mechanisms. It is also important to develop an understanding of the likely contact conditions that exist between the neighbouring components in the belt. A number of the authors discussed in this section propose likely sliding speeds, contact pressures and friction coefficients, that can be used to help in the development of the belt models discussed in chapter 8. The CVT fluid has to perform many tasks in the transmission, from lubrication and heat removal to a signal transfer mechanism in the hydro-mechanical controller. It is also a critical component of the belt mechanism, having to lubricate the assembly but at times permit a high coefficient of friction through which the torque may be transmitted.
Fewkes et al. [43] produced a study of the function of the lubricant in a CVT. In particular the study aimed to investigate the phenomena of ‘scratch noise’ associated with the original derivatives of the Van Doorne belt. The authors predict levels of pressure and relative velocities for a number of contact points within the variator, as tabulated in Table 2-1. It can be seen from these predictions that a vast range of different operating conditions exist between locations and the conditions that occur at any one location may vary considerably as well. The authors tested a number of different fluids under lab test conditions and are able to predict from these results the suitability of the fluid to prevent the ‘scratch noise’ phenomena. In general the stick slip characteristic of the CVT fluid needs to be one where the friction coefficient of the fluid increases with sliding speed rather than decreasing.
Ishikawa et al. [44] studied the effects of CVT fluid on the friction coefficients between metal components. The authors performed experimental friction test methods using a block-on-ring wear test machine to simulate the friction between the belt and pulley on the transmission. The effect of oil detergent additives was also investigated; it was found that the additives deposit a film of 80-90 nm on the sliding surfaces, which acts to increase the friction coefficient between the two surfaces. Two additives were tested, a calcium (Ca) detergent and a zinc dialkyldithiophosphate (ZDDP); both additives were shown to increase the coefficient of friction between two sliding surfaces. The friction coefficient was also shown to decrease again as the additive levels began to deteriorate
Similarly, Ichihashi et al. [45] have also investigated CVT fluid friction properties; they have simulated the belt pulley contact using a test rig applying a 100 MPa load at a sliding speed of 0.1m/s, based on the values proposed by Fewkes et al. [43]. The authors then investigated the effects of blending several different additives with the base oil. Generally calcium containing detergents gave improved friction coefficients, and reduced noise problems, while amide and ester based additives reduced the friction coefficient. However, the calcium detergents had a detrimental effect on the clutch judder tests. It was concluded that the best results were achieved using a blend of both calcium detergents, for improved traction, and phosphate detergents to improve clutch judder performance. Measured coefficients of friction ranged from approximately 0.09 to 0.135. All the fluids tested displayed an increase in friction coefficient, by as much as 30%, as the fluid temperature was increased.
Peiffer et al. [46] describe a range of methods developed by Shell for testing that different blends of CVT fluids have the correct performance characteristics. The paper discusses the conflicting requirements, of a CVT lubricant and how these requirements affect the efficiency of the variator mechanism itself.
The authors note that CVT transmissions need to achieve a wide acceptance with the end user, the driver, and in order to do this they must achieve these performance characteristics with the same standard as would be expected of a manual or automatic transmission. All CVT fluids must therefore have a fill for life performance similar to all modern manual and automatic transmission fluids. The CVT fluids also have to perform many tasks in the transmission, allowing judder free clutch performance, good low temperature pumpability and starting performance, low losses in gears and bearings, high pump efficiency, and high variator efficiency. High pump efficiency requires the fluid to have almost constant viscosity/temperature behaviour, and excellent air release properties, so that the pumped volume does not change due to excessive leakage or changes in the fluid density. The variator efficiency demands on the fluid are more contradictory; a fluid capable of giving high metal to metal friction characteristics allows the clamping pressure to be reduced, thus reducing the pump losses, but at the same time this may increase the power loss due to friction within the other belt components. The authors have developed a test to approximate the power loss in the belt mechanism by measuring the stable temperature difference between the transmission, run at a set condition, and the ambient temperature. The smaller the temperature difference measured the lower the power loss that exists in the belt. A number of the fluids tested produced friction vs. slip characteristics that increase with sliding speed, a beneficial characteristic in the functioning of the variator.
Similar conclusions and desired fluid characteristics are also proposed by Watts et al. [47]. The authors also note the importance of differentiating between dynamic and static coefficients of friction. A high dynamic coefficient is desirable in CVT applications to transmit the torque at the pulley face. However, fluids which display static coefficients significantly higher than their dynamic coefficients are more likely to exhibit noise-inducing stick-slip behaviour.