Abstract: In order to improve the similarity between freeze-thaw model and the prototype, the concept and calculation method of similarity ratio of freezing index and laboratory correction coefficient are put forward. The loess in Harbin is taken as the experimental material; and the average temperature of this area between the year of 2010 and the year of 2012 is linearly simplified. Based on the differential equation with the heat conduction with phase transition, a freeze-thaw cycle model test is carried out for a specimen with a density of 1.60 g/cm3 and a moisture content of 20 % under the constraint of geometry, temperature and time scale by using the similarity theory. The K value of the laboratory correction coefficient is determined as 1.13 and the similarity ratio of the freezing index CI is determined as 1:102.5 so that another freeze-thaw experiment is conducted after an optimization of the temperature control time in the freeze-thaw stage. By comparison, the maximum frozen depth value of the optimized model is 21.33 cm, which is close to the predetermined model value 20 cm and the difference is 6.65 %; and the optimized freeze-thaw period is reduced from 108.14 h to 84.53 h. Therefore, it is reasonable to optimize the duration time of temperature controlled at different temperature controlling stage by using laboratory correction coefficient K and freezing index similarity ratio CI in freezing-thawing model. The ensured that the freeze depth are more similarity to prototype in the geometrical scale and time scale; at the same time reduced the duration time of a freeze-thaw cycle significantly.

Key words: freeze-thaw, simulation test, freezing index similarity ratio, laboratory correction factor