The decision to use a particular separation technique industrially is based on multiple factors. To speak generally, this comes down to ease of separation and process economics. The simplest and cheapest techniques would be ones like biphasic separation or filtration.
The distillation you refer to bears more relation to bulk chemical manufacture such as oil refining. Process HPLC and simulated moving bed (SMB) are more involved techniques, requiring more sophisticated technology and hence, is a more expensive technique. This puts it into the realms of higher value products which would tend to be pharmaceuticals or ultra high purity display chemicals.
So, for process economics, if we think in terms of Input-Process-Output we have cost of input material, the cost of processing and the cost of output material. If Input + Processing are greater than the ultimate commercial cost that the product can bear, then you don't have a product or it would need a key differentiator (which could be purity), to justify the extra cost.
If the costs justify it, then using higher vacuum can offset degradation risk involved in a fractional distillation. Physicochemical factors come into play to select this technique (under the ease of separation banner). Industrial distillation would tend to be used for more complex mixtures of desired volatile and semivolatile components. A good Quora discussion on the topic is here:
The technique, obviously relies on boiling point and exploitation of potential synergies of compound (usually liquid/oil) mixtures.
Chromatographic separation relies on adsorption/desorption by judicious selection of stationary phase and solvent (mobile phase). For commercial processes, this needs to be repeated over and over in a completely reproducible way with minimal variation. If the stationary phase is cheap, then less material has to go through the material before being changed. If the stationary phase is expensive (decent spherical functionalise silica can easily be over £2000/kg, chiral phases over 10-20k per kg. Irregular silica can be £500/kg), then you want a faster, cleaner separation with minimal fouling of the support. Over 200 injections onto the same stationary phase is not unheard of.
These injections would also be maximised on production scale vs the standard analytical technique.
For SMB, typically four HPLC columns are used and a more sophisticated algorithm is required to inject the right amount of material onto the correct column at the right time.
As commented on, isomeric separation, particularly if the materials are solids, then favours process HPLC rather than high vacuum distillation. But again, the process economics would help decide this.
Other factors are internal expertise in the relevant techniques and/or capability of outsourcing partners (contract manufacturing organisations) in more niche techniques. SMB would probably fall into this category i.e. most manufacturing companies would look to a specialist 3rd party unless they were already engaged with this technology.