Improved accuracy of on-line heavy water measurement using infrared spectroscopy by investigation of signal-to-noise ratio and temperature influences

Abstract

Experimental protocols for the on-line measurement of heavy water concentration in nuclear power plants have been established, and important parameters, such as the temperature and signal-to-noise ratio, which govern the accuracy of measurement, have been studied. The temperature of a sample should be controlled tightly because the temperature variation introduces non-linear baseline variations and leads to an increase of the partial least squares calibration error. Furthermore, the variation in the signal-to-noise ratio of spectra sensitively influences the calibration. For reliable prediction, it is critical to maintain the signal-to-noise ratio at a certain level. When the sample spectra were collected at a higher temperature, it was possible to acquire spectra with an improved signal-to-noise ratio and better calibration. In addition, a single beam spectrum of water shifts to a lower frequency, and the maximum transmission intensity at around 2500 cm⁻¹ (the heavy water band location) increases at an elevated temperature. Overall, an on-line infrared spectroscopic scheme is presented for measuring heavy water. The scheme can be applied to an actual process without practical difficulties. If the spectra could be collected at elevated temperature over 2 min with the use of a high throughput light source, the prediction error could reach to 1.0 ppm.