Fifty Years of Progress in Modelling the Accumulation of Atmospheric Ice on Power Network Equipment

The twentieth century witnessed two stages in the development of models to estimate the accumulation of ice on power network equipment. The first stage consisted of analytical models, expressed as simple equations, which could be readily solved by hand or with a calculator. In order to achieve simplicity, these models relied on strong assumptions or constraints about the nature of the icing process. Some of them also incorporated experimentally based empiricisms. For all of them, the goal was to predict bulk ice accretion properties such as ice load. They did not include details of the icing process. The second stage consisted of simple numerical models. It began around 1980 and was pioneered by the present authors. New models were developed which endeavoured to account for the physical details of the icing process. These models required personal computers for their implementation. Nevertheless, numerous assumptions and simplifications were still made in order to keep the computational problem tractable. With the advent of ever increasing computational power, the twenty-first century has seen the development of yet another generation of models which we may call "supercomputer models". Some of these numerical models are by far the most complete in terms of accounting for the physical processes of icing. Others adopt a radically different approach to icing simulation. However, these exciting developments present a serious problem. The quality controlled field and laboratory data that are needed to verify these third stage models are scarce or non-existent. In this paper, we document the progress of ice accretion modelling for power network equipment and reflect on the need for future model development and testing.