Method for predicting the amount of accessible carbohydrate in a feedstock sample using a near-infrared model
The present invention is directed to a method of using a near-infrared (NIR) model to predict the amount of accessible carbohydrate material in a feedstock sample. The NIR model is generated by subjecting a set of biomass calibration feedstock samples to near infrared spectroscopy which produces NIR spectroscopic data, measuring the chemical compositions of each said calibration feedstock sample by wet chemical techniques, and measuring the total amount of monosaccharides and disaccharides solubilized by each said calibration feedstock sample after processing under a defined pretreatment condition or defined enzyme load condition. The method is then applied to NIR spectral data obtained from a test feedstock sample to predict the amount of accessible carbohydrate material in the test feedstock sample.
1. A method of using a NIR model to predict the amount of accessible carbohydrate material in a feedstock sample, comprising:
a) collecting NIR spectral data from a test feedstock sample;
b) applying, to said spectral data, a NIR model of accessible carbohydrate calibrated for feedstocks of the same type as said sample, wherein said NIR model is formulated by: i) subjecting a set of biomass calibration feedstock samples to near infrared spectroscopy to produce NIR spectroscopic data from each said sample, each said calibration sample of the same type of feedstock as said test sample; ii) measuring the chemical compositions of each said calibration feedstock sample by wet chemical techniques; iii) measuring the total amount of monosaccharides and disaccharides solubilized by each said calibration feedstock sample after processing under a defined pretreatment condition or defined enzyme load condition; and iv) generating said calibrated NIR model for predicting the amount of accessible carbohydrate in said test feedstock sample processed under said defined pretreatment condition or defined enzyme load condition, from said spectroscopic data, said chemical compositions, and said solubilized mono and disaccharide amounts; and
c) predicting the amount of accessible carbohydrate material in said test feedstock sample, based on the results of said applying step, wherein steps b) and c) are performed on a suitably programmed computer.
2. The method of claim 1 , wherein said generating step comprises regressing said spectral data against said chemical composition data and said monosaccharide and disaccharide amounts.
3. The method of claim 1 , wherein said defined condition is a defined pretreatment condition.
4. The method of claim 1 , wherein said defined condition is a defined enzyme load condition.
5. The method of claim 1 , wherein: step iii) comprises measuring the total amount of monosaccharides and disaccharides solubilized by each said calibration feedstock sample after processing under a defined pretreatment condition and a defined enzyme load condition; and step iv) comprises generating said calibrated NIR model for predicting the amount of accessible carbohydrate in said test feedstock sample processed under said defined pretreatment condition and defined enzyme load condition, from said spectroscopic data, said chemical compositions, and said solubilized mono and disaccharide amounts.
6. The method of claim 1 , wherein said test feedstock sample comprises biomass from a species selected from the group consisting of switchgrass, sorghum, sugarcane, miscanthus, poplar, willow, rice, wheat and corn.
7. The method of claim 6 , wherein said test feedstock sample comprises biomass from sorghum.
8. The method of claim 6 , wherein said test feedstock sample comprises biomass from sugarcane.
9. The method of claim 6 , wherein said test feedstock sample comprises biomass from switchgrass.
10. The method of claim 6 , wherein said test feedstock sample comprises biomass from corn.
11. The method of claim 6 , wherein said test feedstock sample comprises biomass from wheat.
12. The method of claim 1 , wherein said NIR model is further for predicting the saccharification efficiency of said test feedstock sample.
13. The method of claim 1 , wherein said NIR model is further for predicting the product yield of said test feedstock sample.
14. The method of claim 13 , wherein said product yield is ethanol yield.
15. The method of claim 14 , wherein said test feedstock sample comprises sorghum, switchgrass or wheat.