| Summary: | Accurate measurement of temperature in flames is a challenging problem that has been successfully addressed by hyperspectral imaging. This technique is able to provide maps of not only temperature T (K) but also of column density Q (ppm<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mspace width="0.166667em"></mspace><mi>·</mi><mspace width="0.166667em"></mspace></mrow></semantics></math></inline-formula>m) of the main chemical species. Industrial applications, however, require cheaper instrumentation and faster and simpler data analysis. In this work, the feasibility and performance of multispectral imaging for the retrieval of T and Q<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mrow></mrow><msub><mrow><mi mathvariant="normal">C</mi><mi mathvariant="normal">O</mi></mrow><mn>2</mn></msub></msub></semantics></math></inline-formula> in flames are studied. Both the hyperspectral and multispectral measurement methods are described and applied to a standard flame, with known T and Q<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msub><mrow></mrow><msub><mrow><mi mathvariant="normal">C</mi><mi mathvariant="normal">O</mi></mrow><mn>2</mn></msub></msub></semantics></math></inline-formula>, and to an ordinary Bunsen flame. Hyperspectral results, based on emission spectra with <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>0.5</mn></mrow></semantics></math></inline-formula> cm<inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mrow></mrow><mrow><mo>−</mo><mn>1</mn></mrow></msup></semantics></math></inline-formula> resolution, were found in previous works to be highly accurate, and are thus considered as the ground truth to compare with multispectral measurements of a mid-IR camera (3 to 5 <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi mathvariant="sans-serif">μ</mi></semantics></math></inline-formula>m) with a six interference filter wheel. Maps of T and Q obtained by both methods show that, for regions with T <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mo>≳</mo><mn>1300</mn></mrow></semantics></math></inline-formula> K, the average of relative errors in multispectral measurements is ∼5% for T (and can be reduced to ∼2.5% with a correction based on a linear regression) and ∼20% for Q. Results obtained with four filters are very similar; results with two filters are also similar for T but worse for Q.
|
|---|