20 Minutes With... Dr. Dong Zhu

Tribology & Lubrication Technology, Nov 2007 by Phipps, Karl M

For sliding wear in gears, it is similar that currently used wear models are typically in the form of a simple empirical formula with a few parameters. Most models of this kind were developed long ago for the purpose of rough estimate of wear with manual calculation tools. They do not seem able to describe very complicated wear phenomena and mutual dependence between the wear and mixed lubrication characteristics.

Today new wear simulation and prediction models need to be developed using advanced computer technologies. A collaborative effort has been made with professor Y.Z. Hu's group at Tsinghua University and professor lane Wang's group at Northwestern University to develop a deterministic sliding wear simulation approach based on an available mixed lubrication model. It can predict contact severity at each asperity contact and simulate material removal at local spots where material failure criteria are met. The modified surface profile, after material removal at current time step, will be used for the next time step and surface evolution due to wear that can then be simulated step-by-step.1 Concurrent efforts with similar approaches have also been made for wear simulation by different researchers such as Dr. Daniel Nelias at INSA Lyons in France and Dr. Diann Hua et al. at Caterpillar, Inc.. in Peoria, 111. Further efforts, of course, are still needed to integrate this kind of wear simulation and prediction with gear design.

Furthermore, for gear tooth bending fatigue prediction, currently used methods found in the ISO and AGMA standards are still based on a simplified beam-bending theory developed in the 1950s. We have seen that efforts are currently being made by different researchers to integrate fullscale FEM with gear mesh model generation in design packages in order to develop better bending failure models. I am confident that in the near future model-based tribological failure analysis and prediction software packages will be available commercially that will greatly help design engineers to computerize design and analysis, improve product reliability and shorten product development cycle time, just like what we have already achieved for structure strength analysis and prediction.

Can you explain the status of the mixed lubrication study that's applicable to gear tribology?

It is well known that most machine elements operate in the mixed lubrication regime, where both surface contact and hydrodynamic lubrication coexist. For gears, operating conditions are typically severe, as power is transmitted directly through gear tooth contacts where surfaces are usually rough (as making surfaces super-smooth for gears is more difficult than that for many other components). The importance of the mixed EHL study can never be overestimated, as it is fundamental to gear lubrication effectiveness, efficiency, durability and reliability. The key issues in developing mixed EHL models include:

1. How to handle surface roughness. There are basically two approaches: stochastic and deterministic. Stochastic approaches use a small number of statistic parameters to describe the rough surface characteristics and their influences on contact and lubrication. They predict global performance and average values, but localized details and peak values (which may be directly correlated to failures) are missed. In the last 10 years, more attention has been given to deterministic approaches due to advancements in computer technologies. Rough surfaces can now be digitized and data matrices used directly in mixed EHL models as input data, so no statistic parameters are necessary (see Figure 2). This type of model can provide complete details for lubricant film thickness and pressure distributions, as well as subsurface stresses. It enables in-depth study on lubrication transition and breakdown, as well as surface failure mechanisms.