Development of an emissivity compensation algorithm for radiometric temperature measurement during the galvanneal process

Lynn Krajnovich Zentner, Purdue University

Abstract

During the galvanneal process, zinc-coated steel sheet is rapidly annealed in order to improve the characteristics of the final product. The zinc layer changes from highly specular liquid zinc with a spectral emissivity of approximately 0.2 to a largely diffuse intermetallic layer with a spectral emissivity as high as 0.8. These rapid emissivity changes preclude standard radiometric measurement techniques. The objectives of the research were to explore the behavior of the spectral emissivity during the process as influenced by the heating cycle and metallurgical properties of the product and to develop an appropriate dual-wavelength emissivity compensation algorithm to accurately infer temperature. The galvanneal process was simulated in the laboratory for 63 samples, and in situ emissivity and temperature data were obtained using a dual-wavelength radiation thermometer at spectral bands of 2.18 and 2.4 $\mu$m. Galvanneal temperatures simulated ranged from 753-833 K. Line speeds varied from 60-100 meters per minute. An emissivity compensation algorithm was developed using a mathematical model of the relationship between the two spectral emissivities. A linear relationship between the emissivities produced the most accurate results while maintaining simplicity. The algorithm inferred temperature with a maximum error of less than $\pm$20 K. A modified approach examined splitting the data into three sets corresponding to specific process conditions, each with its own compensation equation, and using these equations in a similar algorithm. This approach performed slightly better than the first, more global approach, but does require knowledge of process conditions in addition to spectral radiance data. A final, completely new approach, utilized an adaptive technique for adjusting the compensation equation within the algorithm according to the behavior of the data being measured. This approach frequently outperformed the previous approaches by as much as fifty percent. Iron analysis results indicated that emissivity increased with increasing iron content. Final iron content varied from 5-13%. Optical and mechanical roughness measurements indicated that surface topography is not the dominant mechanism affecting emissivity changes. Arithmetic roughness values varied from 1-3 $\mu$m. Further research is proposed to explore the effects of layer morphology on emissivity behavior.

Degree

Ph.D.

Advisors

DeWitt, Purdue University.

Subject Area

Mechanical engineering|Metallurgy

COinS