As it was already pointed in the recent post, the friction on ice is an important topic and its typical low value is attributed to the formation of the water film.
Despite our everyday life experience, at low speeds, the ice friction can be quite large. For example, at the surrounding temperature of and sliding speed of the kinetic friction is about 0.6. At the speed of it drops to 0.06. Although the latter value is small, it is still higher than the theoretically obtained value of in case if the uniform film of water is formed between the surfaces, according to Dr. Persson in his recent study. Therefore, he concludes, there exist a mixed state with small ice and water domains, which fluctuates in time and space.
There exist a transition from high kinetic friction of ice to low as a function of speed and at each speed, the friction is determined by the balance of frictional heating and melting and refreezing. Prior to melting of the ice, the friction is high and therefore there is much frictional heating. However, once the melting starts, the friction drops and less heat is provided to the contact. Then, the water has to refreeze. In addition, the process is non-unoform in space due to surface irregularities. Besides the melting, frictional heating results in softening of the ice at temperatures close to the melting temperatures.
Dr. Persson in his articles also presented the model for calculation of the friction coefficient of ice. The model can also be applied to other systems, where phase transition takes place.
More details can be found in the original work by Dr. Persson: Ice friction: Role of non-uniform frictional heating and ice premelting by B. N. J. Persson.