Method and apparatus for real-time measurement of fuel gas compositions and heating values
An exemplary embodiment can be an apparatus for real-time, in situ measurement of gas compositions and heating values. The apparatus includes a near infrared sensor for measuring concentrations of hydrocarbons and carbon dioxide, a mid infrared sensor for measuring concentrations of carbon monoxide and a semiconductor based sensor for measuring concentrations of hydrogen gas. A data processor having a computer program for reducing the effects of cross-sensitivities of the sensors to components other than target components of the sensors is also included. Also provided are corresponding or associated methods for real-time, in situ determination of a composition and heating value of a fuel gas.
1. A method for real-time in situ determination of a composition and heating value of a fuel gas, the method comprising:
measuring an amount of hydrocarbons and carbon dioxide in a fuel gas using a near infrared sensor producing raw measurements of said hydrocarbons and carbon dioxide in said fuel gas;
measuring an amount of carbon monoxide in said fuel gas using a mid infrared sensor producing raw measurements of said carbon monoxide in said fuel gas;
measuring an amount of hydrogen in said fuel gas using a semiconductor based sensor producing raw measurements of said hydrogen in said fuel gas;
collecting all of said raw measurements in real time using a hardware control and data acquisition system;
arranging all of said raw measurements in a first data matrix;
arranging a composition and heating value of said fuel gas in a second data matrix; and
processing said data matrices using a multivariate regression method selected from the group consisting of principal components regression, partial least squares regression, and partial least squares regression 2 , thereby producing models that correlate sensor measurements tocomponent compositions and total heating value of said fuel gas.
2. The method claim 1 , wherein said data matrices are processed using principal components regression.
3. The method of claim 1 , wherein said data matrices are processed using partial least squares regression.