The importance of cavitation in lubrication hydrodynamics is well recognized. Cavitation can also act as a source of experimental error in rheological measurements. Therefore, the ability to understand and predict cavitation is important for tribology. Nearly all models for cavitation prediction are based on the local hydrodynamic pressure. The appropriateness of this approach when viscous stresses are of the order of the hydrodynamic pressure is questionable. One cavitation model that considers the state of stress in a flowing liquid is the principal normal stress cavitation criterion (PNSCC), which proposes that cavitation will occur when the most tensile principal normal stress exceeds some critical value. Although this hypothesis can accommodate many experimental observations, its theoretical foundations are weak. In particular, it fails to account for the tensile strength of liquids and resulting need for nucleation sites; it neglects the role of transport of dissolved gases; and it does not consider the effect of a growing bubble on the local flow, and hence local state of stress. We demonstrate cavitation in low Reynolds number Couette flow, and present a model for cavitation in shear in the limit of creeping (Stokes) flow, which corrects for the theoretical failures of the PNSCC. We use numerical simulation to analyze cavitation onset, and obtain a more general cavitation criteria from which the PNSCC is recovered under certain conditions.
