Do you want to travel from Utrecht to Lyon, from New York to Chicago, or from Mumbai to Bangalore within a reasonable time, cost, and environmental footprint? What about a reliable high-speed train? And what does this have to do with inlet film thicknesses in bearings?
The railway market has two drivers: (1) need for high speeds to go further and faster and (2) need for extended maintenance intervals to increase the network capacities while limiting maintenance operations. With twofold extended maintenance intervals of bearing units, the train operator saves money and divides by two the CO2 emissions associated to the life cycle of each bearing. There are approximately 60 bearing units per train each weighing ~30 kg.
There is a fundamental challenge in combining (1) and (2) which translates into a tribological challenge: with higher speeds, higher power losses, higher operating temperatures, reduced grease life, reduced maintenance interval… To combine (1) and (2), SKF follows three approaches. One of them is to optimize internal geometries in the railway bearing units to reduce friction, and thus operating temperatures. The two others remain a topic for future articles. The optimization of the bearing internal geometries is based on quasi-static and dynamic simulations with SKF proprietary software which includes a friction model prior experimental verification on full-size railway bearings and EN12081 homologation tests. A railway wheel-end bearing unit is typically a grease-lubricated double row bearing unit, with either a tapered or cylindrical configuration (see Figure 1). The operating conditions and sealing solutions used in high speed train applications are such that the cage and seal contributions to the output torque are assumed to be negligible in the friction model.
The simulated output torque is mainly driven by the raceway and flange contacts between the rolling elements and the inner and outer rings. Therefore, the main factors influencing the frictional behaviour are:
- The normal load of the respective contacts, all individually calculated by the model.
- The boundary friction coefficient measured for the main high-speed train greases using a ball-on-disk test rig alternating long mixed lubrication periods and Stribeck curves.
- The surface topography of the respective contacts specified and measured on each bearing components.
- The film thickness in the respective contacts, which are calculated based on:
- the entrainment speed (calculated based on the shaft rotational speed and internal kinematics of the bearing),
- the lubricant viscosity (calculated based on the lubricant rheology and measured operating temperatures),
- the amount of lubricant at the raceway and flange contact inlets which controls the starvation level in the elastohydrodynamically lubricated contacts.
This last point, the inlet film thickness of raceway and flange contacts, respectively hnom,race and hnom,flange are user-defined lubrication parameters. As of today, the inlet film thicknesses cannot be measured in an operating bearing (putting aside a unique test setup in Imperial College London with a small-size ball bearing with a sapphire outer ring and dedicated optics) and it cannot be calculated based on models with reasonable efforts. A common practice when using the SKF model to compare one design to another has been to use a constant value hnom coming from years of experience and a comparison to test results obtained years ago.
Intuitively, one would argue that the inlet film thickness should vary with the bearing type and size, the operating conditions like the temperature, and with the lubricant properties such as the grease bleeding rate. In that sense, for the optimization of internal geometries of railway bearing units to become more predictive than comparative, it is necessary to calibrate the models by adjusting hnom,race and hnom,flange so that the predicted output torque best matches the measured torque on a full-size unit. This is what we achieved: a friction-based calibration of the inlet film thicknesses in railway bearings.
A dedicated torque measurement device has been designed to measure friction on railway bearing units mounted in axel boxes (see Figure 2). It is calibrated with deadweights and can then measure fractions of Nm in torque while transmitting a full radial load of 92 kN on the bearing unit! After a pre-test phase to churn the grease in the bearing and to reach stable operating conditions, torque and operating temperature are measured for various combination of rotational speed and loads, each time being averaged over 30 min both clockwise and counterclockwise.
For each combination of speed and loads, the measured operating temperature is provided as input and a semi-automatic method calculates the output torque of the bearing unit varying independently hnom,race and hnom,flange. Both hnom,race and hnom,flange are assumed to be uniform across the contact width and constant for all the respective raceway and flange contacts. For each combination of speed and load, the respective deviation to the measured torque is plotted and the optimum line of [hnom,race, hnom,flange] points accurately predicting the torque is extracted. Above a certain load and within a relevant range of operating temperature, these optimum lines all cross at an optimum point [hnom,race, hnom,flange]opt for which the predicted torque matches the measured torque. With this optimum point model becomes predictive and can quantify the benefits of internal geometry optimization on friction.
Not only the friction-based model calibration enables more accurate performance prediction of new designs, but it also enables to indirectly assess the inlet film thickness of raceway and flange contacts (to a certain degree of confidence). It reveals for example that for some bearing types, the flange contacts can have two times more lubricant available than the raceway contacts. It also gives insights into how at a given speed, load, and thereby temperature, influence the inlet film thickness. And the method can be used to identify how new grease formulations contribute to lower friction. Is it inside the contacts? Is it at the contact inlet? Is it more in the flanges or on the raceways?
This are the interesting perspectives that come up when experimental and numerical efforts are paired more than just to verify one another. Next time you take a high-speed train, have a look at the bogies. If you get the right micrometric inlet film thickness with the right internal geometry, you will extend the maintenance intervals by a million kilometer and lower the CO2 footprint of your trip.
For further information please contact Dr. Arnaud Ruellan.
Arnaud Ruellan. After a PhD in INSA Lyon and NTN-SNR on early bearing failures and a participation in a Marie Curie European Program in Imperial College London, Arnaud is now Senior Researcher at SKF, Houten, the Netherlands.