Hello!

I would like to know the % error permissible on retention time predicition using mathematical model developed by factorial design in an HPLC analysis?
In a factorial design experiment conducted I am getting an error around 6% on retention time prediction. Is it tolerable? Please clarify this aspect. Thanks for your time.

Depends on what you mean by "6%".

Also depends on what you're doing.

For example, if you are checking hypotheses about chromatographic behavior, then being able to predict retention within +/- 6% (I'm assuming 95% confidence interval here) may be quite acceptable.

If you're trying to use this as a method development tool, you should be looking at 1-2% errors. To put this in perspective, dedicated chromatography modeling programs like DryLab (disclaimer here: I work for LC Resources) typically give predictive accuracy around the 1% level for the simplest models, and this can be improved by using more calibration points.

If the errors for all of the peaks in your separation are correlated (similar magnitude and direction), then your resolution predictions (which is what we really care about) will be better because some of the error will go away when you take the difference in retention times.

Experimental error should be less than 1% (if you're doing well-controlled chromatography).

-- Tom Jupille / LC Resources