Chemometrics is one of the EIRA Ltd. activities mainly related to FOSS Analytical A/S infrared instruments. As every indirect measurement method infrared measurements also works well, when mathematical prediction models are well developed, and works badly, when prediction models are not developed. So, good spectrum analyzer is only half of the story. To finish the story, you need mathematical prediction model, which translates obtained absorbtion or transmission spectra to quantitative parameters you are interested in, like fat, protein, etc.


Ready to use models

Of course for the commercial instruments in major cases those models are already available. So, if you want to analyse let’s say protein in wheat, you have simply to purchase wheat calibration together with analyzer. But in case you want some not so traditional parameter or you produce products with different recipe and/or technology, you can easily get into situation, when mathematical models have to be built from scratch.


Ready-to-use model evaluation

This is the point where EIRA Ltd. can help you. First logical step is to evaluate available ready-to-use models. It means you have to prepare around 20 samples of your product with known reference results. In case you do not have your own chemical laboratory, you can use EiraLab Ltd. services. In the second step we have to obtain spectrum data for your samples with IR spectrum analyzer. In the third step by means of software we are going to model analyzer behavior with different ready-to-use models. In case we get satisfactory results from some model, then we simply adapt it and everything is OK. If none of the ready-to-use models gives satisfactory results, then only choice is the new prediction model(s) development.


New model development

Initialy we will make use of those 20 samples, mentioned before. Of course 20 samples for serious model is too little. But we can start with them and then add new samples when it is possible. It is hard to say how many samples are needed to make a good model. However, to estimate your expenses roughly, you have to assume you will need to make chemical analyses for 50 – 100 samples. Then add EIRA Ltd. prices and you will get approximate total how much it could cost.

If you think this is good investment, we will be glad to cooperate!


Existing model adjustments and corrections

Of course EIRA Ltd. besides with new model development also offers every possible corrections to existing calibrations and models. Here you can find existing calibration correction prices. If you have special case, which is not mentioned in the above document, contact us, please!


Calibrations made by EIRA Ltd.

Additional parameters for Oats for Infratec analyzer

Fat DM Starch DM Beta glucane DM B-glucane in naked oats
Number of samples 204 150 122 27
Range 3,7…7,9% 43,2…52,3% 2,5…4,8% 3,5…4,8%
Mean/SD 3,32/0,39
SEC 0,38% 1,00% 0,28% 0,24%
SECV 0,39%
SEV 0,36%
RSQ 0,901 0,836 0,49*
1-VR 0,16*

* – parameter is not objective because of the limited range

Traditional parameters like moisture and protein are included in FOSS A/S calibrations.

Rapeseed cake for Infratec analyzer

Moisture Protein DM Fat DM Fibre DM
Number of samples 79 79 79 51
Range 6,6…12,4% 30,5…36,0% 5,5…23,2% 9,9…17,5%
SEC 0,21% 0,24% 0,22% 0,46%
SECV 0,22% 0,29% 0,26% 0,47%

Barley calibration for InfraXact analyzer

Lysine DM Protein DM
Number of samples 138 91
Range 0,35…0,59% 9,9…19,2%
SEC 0,011% tiek testēts
SECV 0,020% tiek testēts

Rapeseed cake calibration for InfraXact analyzer

Moisture Protein DM Fat DM Fibre DM
Number of samples 115 107 104 85
Range 3,5…11,8% 26,0…40,8% 0,9…33,2% 5,8…19,8%
SEC 0,35% 0,36% 0,24% 0,45%
SECV 0,43% 1,05% 0,41% 0,70%