A hypothesis that large-scale fracture of sea ice plates in the Arctic could be caused by the release of energy of thermal bending moments due to major temperature changes is advanced and examined. Bending propagation of a through-the-thickness crack along the floating plate, with negligible inertial forces, is analyzed, assuming the moment field in the plate near the traveling crack front and the fracture process zone to be in a steady state. The analysis uses the plate-bending theory, and the second-order geometric effects of the in-plane normal forces are taken into account. Quasi-elastic behavior is assumed, and creep is treated approximately according to the effective modulus method.

The calculated temperature difference between the top and bottom of the plate required to produce this kind of fracture is found to be well within the range that actually occurs in the Arctic, but this cannot be regarded as a proof of the hypothesis because of the simplifying assumptions made as well as uncertainties about large-scale fracture properties of sea ice. Further, it is shown that this type of fracture must exhibit a size effect, such that the critical temperature difference decreases in proportion to (plate thickness)-3/8. This might explain why large fractures often form in an intact thick plate rather than only in a thin plate and along lines of weakness. For the case that the in-plane forces are significant, it is shown that beyond a certain critical crack length the thermally driven bending fracture (if it exists) must transit to a planar (nonflexural) fracture driven by the release of the energy of the in-plane forces generated by wind and ocean currents. The effect of creep is to increase the required critical temperature difference, as well as the critical crack length for the aforementioned transition. For thermal bending fracture, the minimum possible spacing of parallel cracks increases with the plate thickness and is independent of the crack length, while after transition to planar fracture it increases in proportion to the crack length. The hypothesis of thermal bending fracture cannot be proven or disproven without new types of experiments and measurements in the Arctic, and their computer modeling. ¿ American Geophysical Union 1992