New computational tool to evaluate experimental VLE and VLLE data of multicomponent systems

•A computational approach is performed to validate VLE and VLLE data.•Application on multicomponent systems qualifies the individual and global errors.•Modeling results affect directly to the data consistency test.

This work presents a rigorous method to analyze the thermodynamic consistency of VLE and VLLE data of multicomponent systems, as an extension of a method previously proposed for binary solutions. The method or proposed test verifies the coherence between the Gibbs-Duhem equation and experimental data, using the same number of equations as degrees of freedom of the system. Resolution is achieved by means of two methods called the integral-form and differential-form, and for each of these, characteristic parameters that qualify the quality of the data are generated. The integration of the equation above mentioned between data pairs generates the residuals δψ and constitutes the integral-form. This form verifies the consistency when the value of these residuals is lower than the maximum value, calculated as εψM=κψεψM,0, where εψM,0 is the error associated with ψ at each point and κψ = 5; it should occurs that εψM ≤ 4. In the application of the differential-form each partial derivative of ψ is verified and can be used to verify the coherence between the compositions of each component in all the phases by the residual δζi. The maximum values of these residuals are established by εζiM=κζiεζiM,0, where εζiM,0 is the maximum permissible error and κζi=5, it should occurs that εζiM≤|0.1[max(ζi)-min(ζi)]|. The limits of the parameters for the proposed test are established after applying the method to several systems generated artificially. The test was applied to a set of real systems, 50 ternaries and 2 quaternaries, verifying the degree of consistency/inconsistency according to the parameters defined. The behavior of the test is compared with that of Wisniak-Tamir in multicomponent systems. In summary, the proposed test is shown to be a useful tool to assess the quality of VLE and VLLE data of multicomponent systems.