A systematic investigation on the accuracy of computer simulations for optical isomers in industrial settings

• Predicting the band profiles of typical chromatographic separations using the approach of the inverse method (IM) [2] requires finding the numerical solution of the Equilibrium-Dispersive (ED) mass balance equation [1]. The accuracy of the determination of the competitive adsorption isotherms is therefore explicitly dependent on the choice of algorithm to solve the ED model. Earlier studies have shown that the well known RFD algorithm [3] has limitations in its accuracy for certain simulated conditions such as column efficiency and adsorption isotherm type [5]. In this study, we thoroughly investigate three different algorithms, the rapid and well known Rouchon Finite Difference (RFD) algorithm [3], the Orthogonal Collocation on Finite Elements (OCFE) method [4] and the Central Difference Method (CDM) algorithm developed by us. Firstly, a systematic investigation is made, comparing the different algorithms under a broad range of different synthetic conditions. Secondly, we apply all three algorithms using the IM on real experimental systems; (I) one high efficiency quaternary separation and (II) one pharmaceutical industry application with a chiral intermediate separation. We have found discrepancies between the conventional (RFD) and the more accurate (OCFE and CDM) algorithms for several synthetic conditions such as low efficiency, increasing number of simulated components and components present at different concentrations. Given high enough efficiency, we experimentally demonstrate good prediction of experimental data of a quaternary separation problem using either algorithm but better prediction of OCFE/CDM for a binary low efficiency separation problem. Our conclusion is to use the RFD algorithm with caution when any of the particular conditions we have investigated is valid. When the highest accuracy is sought, there is no doubt that OCFE or CDM should be the algorithms of choice. However, given the computational speed of RFD, we also recommend it for preliminary parameter fitting. For conditions where we have shown the algorithms to produce practically identical solutions, RFD can be applied with greater certainty. It remains to be thoroughly investigated the effect on the particular shape of the adsorption isotherm on the obtained solution, for example type III having inflection points.

Ämnesord

NATURVETENSKAP  -- Kemi (hsv//swe)

NATURAL SCIENCES  -- Chemical Sciences (hsv//eng)

Nyckelord

Kemi

Chemistry

Publikations- och innehållstyp

ref (ämneskategori)

kon (ämneskategori)