The wide acceptance of the plate-tectonics model to describe inferred large-scale motions of the earth at the surface has prompted considerable activity in attempts to construct models of circulation of the mantle. These models must be consistent not only with the observations of surface motions, which are kinematic constraints, but also with certain rheological and geophysical constraints.
Additionally, the driving mechanism must be sufficiently potent to overcome the energy losses in the system which are, minimally, the friction in the circulation as a result of earthquakes and viscous-drag forces. To date, only processes of thermally driven convection seem to be energetically potent. The source of the heat has been variously proposed to be one or more of the following: the core, radioactivity in the mantle, differentiation of mantle and crustal material, and differentiation of radioactivity into the crust. Deceleration of the earth's rotation is only marginally potent enough to generate plate motions.
The choice of rheologies which have been proposed ranges from those for materials with finite strength, and with properties of brittle fracture, to those which exhibit constant viscosities under linear stress-strain relations. Undoubtedly, both these extremes are inappropriate to the entire mantle. In the case of the first extreme, it has been shown that convective motions are not possible if the mantle rheology is similar to that of materials under-going brittle fracture at the surface of the earth. However, this rheology is inappropriate because the influence of increasing temperature and pressure weakens earth materials. A more appropriate rheology