In industrial liquid separation processes chromatography often has a key function in the optimization of yield and purity. For the design of an industrial system, chromatographic processes are generally simulated using mathematical models, tested and optimized at laboratory level, and then scaled up to pilot and subsequently industrial scale. To describe the system, experimental data and model data need to be fitted and extra column contribution must be determined. This paper describes the influence of extra-column volume on overall separation efficiency for lab scale and its impact on the design of large scale systems.
Measurement of extra-column contribution was investigated in terms of mean retention time and variance using two different methods the commonly used zero dead volume connector and as an alternative the zero length column. Further a technique is presented to estimate extra-column contribution to band broadening for different injection volumes, velocities, and tracers based on representative measurements.
When scaling up, often contribution of extra-column volume from laboratory equipment is neglected assuming to be on the safe side, however column efficiency is often lower than efficiency measured for the entire chromatographic system. Relation between system efficiency and column efficiency was investigated using laboratory data and the lumped kinetic model. Depending on the ratio of extra-column volume to retention volume in the system, deduced column efficiency was up to 20% smaller than overall system efficiency. This ratio revealed the misleading nature of the term efficiency loss, when describing influence of extra-column volume on column efficiency. A scheme, which relates the relative variance of the system to the relative extra-column volume, provided an assessment of under- or overestimation of column efficiency. In this article it is shown how scaling up a system based on laboratory data, where extra-column volume contribution is not accounted for, may severely overestimate column efficiency. This overestimation results in underestimated column dimensions at pilot and industrial scale, and hence underperformance of the industrial system.
