Abstract: To quantitatively evaluate the partial similarity modelsof water flooding, a new numerical approach of sensitivity analysis isproposed for the sensitivity of the relaxed dimensionless parameter.The sensitivity factor quantifying the dominance degree of the relaxeddimensionless parameter is defined. By solving the dimensionless governingequations including all dimensionless parameters, the sensitivity factor ofeach relaxed dimensionless parameter is calculated for each partialsimilarity model, thus the dominance degree of the relaxed one is quantitativelydetermined. Based on the sensitivity analysis, the influence coefficientof thepartial similarity model is defined as a criterion to evaluate each partialsimilarity model. The partial similarity model with the smallestinfluence coefficient can be singled out. Results show that theprecision of the partialsimilarity model is not only determined by the number ofdimensionless parameters but also by the relative relaxation quantity.