The model of liquid chromatography with a stationary phase (silica, alumina, ...) and a mobile phase (your eluent / eluent mixture) is a simplification. If your liquid chromatography is not based on size exclusion, then recurrent adsorption of molecules to, and desorption of molecules from the stationary phase are detrimental to the success of your intended purification.
Note, this back and forth isn't only about the molecules you want to separate. It equally happens to the molecules of your eluent (which itself can be a mixture), which are of much higher concentration. Hence, the dissipation of heat of adsorption can become an issue, especially for a sudden substitution of one eluent (mixture) by one other, or if the gradient to change the eluent's composition is too steep: The temperature can rise this much that outer walls of the (preparative) column feels warmer, and that solvents of lower boiling point pass this threshold. A preparative column filled by gravity / settlement of a suspension then can break. Perhaps someone else can back my impression with experimental data, the larger the ratio of diameter to enclosed volume of a preparative column, the more likely this is to happen.
(A rise of temperature equally can be observed preparing the silica suspension if solvent is added to still dry silica.)
VWR / Ace glass offers jacketed chromatography columns with dimensions suitable for preparative scale; in principle, their connection with a thermostat could attenuate the problem. So far, I didn't use them; their threads suggests they might anyway be designed for automated use with a fraction collector, or a MPLC:
(picture credit VWR)