Correlation regression methods for heat exchanger have been developed in last several years

such as the method based on equal flow Reynolds number

the method based on equal flow velocity

and the Wilson plot technique

improvements of which contribute outstandingly in measuring heat transfer equipment. The defect in those methods is as follows. The methods based on equal flow Reynolds number and flow velocity restrict the flow conditions

duct structure and flow media in two ducts. The Wilson plot technique demands the exponent of Reynolds number. A new method is presented in this paper

which is similar to the Wilson plot technique but avoid above restrictions through searching proper coefficient and exponent in the correlation by examining the minimum variance in different fitting consequence

giving accurate heat transfer coefficients on the fixed side. The relation between two ducts is the thermal resistances in series. The method resolves the correlation fitting technique with a dichotomy

which is more accurate than linear regression analysis commonly used. The method is validated by simulated data first and then by experimental data. The result shows that the method is reasonable.