for the prep/semi prep HPLC purification, i usually developed analytical method on 4.6mm C18 column then scale up to prep column like 30mm or 50mm diameter same type C18 column(i mean exactly same brand and manufacturer). the final method is usually isocratic since there are always some impurities close to target peak in the initial gradient run. (this makes the peak broader than in initial gradient run).
my question is:
the calculation shows that in most of the cases, the injected amount onto a prep C18 column is only about 2-5 mg by using calculated scale-up of developed analytical method to get a acceptable peak purity(i always tried to overload it), though the chromatogram is fully similar to the analytical data. this is well below the suggested amount from manufacturer(waters says in its catalog that it can load 200 mg per injection.) so i wonder if this is normal, or there is a better way to reach the supposed 200mg loadability? i doubt this depends on the specific phase material or not? the sample usually dissolved in ACN or ETOH at about 20mg/ml max. concentration.
i know there are many factors in this, but shouldn't it be around one order of difference? i mean if the load capacity from the table is 200mg, then it should at least load 20mg in practical, right? a few mg is too low.

so how to define the scalability of analytical method here for prep purpose?

There are a multitude of factors that can limit loadability. You appear to have at least one problem: your sample is dissolved in an organic solvent. Under these circumstances, you are not limited by the sample load on the column, but by the fact that you are injecting an organic solvent. This results in a range of problems, from early breakthrough of the sample to peak distortion at low load.
You can either attempt to dissolve your sample in mobile phase or a solvent composition that is a weaker eluent than the mobile phase. Or you can work with a technique that we have called at-column dilution. We got an article in print that describes the technique.
Also, it is possible that this is only part of the problem. What is the nature of your sample? Is it ionic? Can you adjust the pH to make it non-ionic? We have found that you can get a factor of 50 higher loadability for the same sample if it is in a non-ionic form.