Numerical estimation of multicomponent adsorption isotherms in preparative chromatography: implications of experimental error.

Since experimental methods for measuring multicomponent adsorption isotherms are extremely tedious, numerical approaches are an attractive alternative. Here, the variance in isotherm parameters as a function of experimental error in measured effluent concentrations is quantified. The number of experimental replicates needed to obtain isotherm parameters to a desired level of accuracy is calculated explicitly. After the covariance matrix of the parameters has been determined, Monte Carlo methods are found to be rapid and effective. The use of different kinds of experiments, the effect of resolution and loading, and the impact of the number of measured data points are described.