Excess partial molar enthalpies of binary liquid mixtures can be accurately measured at very low concentrations by means of continuous-injection heat conduction calorimetry. This method causes evolution of the thermal properties of the calorimeter as a function of time. In this paper it is shown that identification and deconvolution of the thermal signal are equivalent problems when dealing with time-varying systems. A signal treatment analysis is proposed based upon optimal tracking, and different numerical examples, where the actual calorimeter is replaced by a discrete RC model, are presented.

A general deconvolutive method developed from Optimal Control Theory is presented and applied to flux or conduction calorimetry. It requires a previous identification of the system and gives very similar results to equivalent inverse filters (i.e. using the same model). Nevertheless, on the one hand the thermogenesis has to be obtained off-line (inverse filters allow on-line deconvolution) but on the other, the method can handle time-varying systems.