Classification, deconstruction and evaluation of frost heave models: How modeling methods cause simulation error

• Modeling methods of frost heave models for the past 50 years were deconstructed. • Error sources of each coupling components of the models were analyzed. • Regarding freezing soil as porous medium causes water migration error. • Error sources of the model can be hidden by coupling components error. Frost heave models are important tools to study engineering damage in cold regions. However, multi-physical-field-coupling and different methods of coupling components bring difficulties to compare the models and to optimize their error sources. Therefore, classifying and deconstructing of different frost heave models and further evaluating them are significant for model development. We reviewed frost heave models for the past 50 years and built frost heave model families to show the classification and inheritance relationships. Then, the models were deconstructed by the thermal–hydraulic framework (THF) and frost heave quantification method (FHQM). Finally, based on simulated and experimental data in cited papers, the model errors caused by different THFs and FHQMs were analyzed. The results show: (a) Present frost heave models can be classified as porosity increase based (P) and ice lens increase based (I) models and they are tending to merge with each other; (b) THFs of P models regarding freezing soil as porous medium are the main reason for water migration error in the region of multi-ice layers, which also results high uncertainty of hydraulic parameters under different simulation conditions; (c) The most used two FHQMs of P models cause error exceeding 20 % because inaccurate relationship between pore strain and frost heave displacement; (d) The errors of THFs and FHQMs offset each other, which will hidden the error sources of the models. This paper can provide a reference for the improvement of frost heave models used in engineering.