Recently,I have visited an ethane thermal cracker pilot plant. During the process of discussions I found that the product composition reported is much lower than expected for the process conditions maintained. We observed that two separate GCs are used to analyse hydrocarbons and hydrogen. External standard method is employed for quantitation using calibration gases imported from M/s Spantech, UK. After the analysis the total of all the component mole percentages is coming to some value between 60 to 90 percent. Product is not being analysed for CO & CO2 every time. Their total is below 1 percent and so neglected for routine analysis. Then individual compositions are normalised to get a total of 100 percent. Analysts say that what they are doing is right. I find in a book by Cowper and DeRose of British Gas Corporation on gas chromatography that normalisation is to be done only for cases with total value close to 100%.
It is also observed that reported values for ethylene and hydrogen do not meet stoichiometric requirements of the reactions taking place. I am a process engineer with very little knowledge about chromatographic analysis.
I shall be grateful to get any details on normalisation for quantitaive analysis purpose.
Can the results of analysis from two different columns be combined for normalisation?
If the composition data is to be normalised everytime a sample is injected, what is the purpose of calibration?
Is it right to normalise the data for a total of 60%?
In anticipation of valuable suggestions.
K. Babu Rao
By Anonymous on Thursday, August 21, 2003 - 04:58 am:
Hello K. Babu Rao,
the advantage of normalization is the difference in injection volume is minimized. With gas analysis, the injection is normally done with a gas sampling valve and a sample loop.
The pressure in the loop has a direct influence on the amount of sample injected (according to the gas law). Therefore, if the pressure in the loop is slightly higher than used for calibration, you can find "recoveries" about 100%. Of course, the opposite also is true. This might explain why the total of 60% is found for the samples.
Another option is that not all components are quantified. Like mentioned, if you do not take into account the Co and CO2, this concentration should normally be subtracted from the 100%. If itís only a trend you want to follow, the absolute numbers might not be interesting and you can leave them out of routine analysis.
The calibration has also a purpose to calculate the response of the detector. This response can vary slightly in time for the individual components due to changing settings.
These factors are used before the normalization.
Normalization of 60% recovery is not quite usual. Normally it is accepted to normalize if the recovery lies within 98-102%. This values are depending on the lab policies, but it gives you an indication what to look for.
Personally, I think recoveries of 60% mean there is something wrong with either the injection of the calibration analysis or the analysis self. We normally advise to do a atmospheric injection of gasses, because the atmospheric pressure is quite stable.
I hope this gives you some helpful information.
With kind regards,
AC Analytical Controls