(193a) A Constraint on Classical Least Squares for Chemometric Monitoring in Multicomponent Nuclear Waste

Advances in spectroscopic equipment and analysis techniques (chemometrics) of vibrational spectroscopy have enabled the use of process analytical technology (PAT) in a variety of applications. A situation often encountered when applying spectroscopy to a continuous monitoring task is that unknown species, or adulterants, can disrupt quantification predicted by PAT. The disruptive behavior of unknown species can lead to poor model robustness and prediction errors, which can lead to misinformed or delayed process decisions. Many methods exist for identifying and removing additional sources of variation that are not included in calibration data. However, existing methods either require estimates of the pure-component spectra of unknown species or rely on multiple time-series measurements to be effective. The method discussed here, a constraint on established Classical Least Squares methods, avoids the requirement of time-series data and the need for pure-component spectra of unknown species, while still proving effective in the case of poorly resolved or overlapping peaks. The applied constraint is that unknown species will have a nonnegative spectral contribution. All species obeying the Beer-Lambert law cannot physically have negative spectral contributions. A constrained Classical Least Squares problem produces a more physically realistic solution than an unconstrained Classical Least Squares solution in cases where spectra closely follow the linearity of the Beer-Lambert Law and unknown species are present. This work focuses on applying this constrained Classical Least Squares approach to multicomponent streams encountered at the Hanford Site in Washington State.

The effective utilization of data has become more important as processes are providing more information with Industry 4.0 initiatives [1]. One application where data utilization has the potential for process improvements is the Hanford nuclear waste vitrification process plant in Washington State. This site removes water and radionuclides from radioactive waste before vitrifying the waste for stable long-term storage. However, the vitrification process is expected to have inconsistent feed stream compositions, leading to the need for real-time process monitoring. A constrained Classical Least Squares approach has the potential to strengthen the robustness of real-time chemometric models monitoring the Hanford site.

Two on-line sensors are investigated in this study: Raman spectroscopy and Attenuated Total Reflectance – Fourier Transform Infrared (ATR-FTIR) spectroscopy. Both are vibrational spectroscopies that can benefit from a constrained Classical Least Squares approach to quantification in complex mixtures. ATR-FTIR derives its linearity from the Beer-Lambert Law, while Raman scattering is analogously linear under certain conditions [2]. Using physical relationships that suggest linearity is the theoretical underpinning of quantification models such as Classical Least Squares. The unknown or adulterating species are also assumed to obey the Beer-Lambert Law and to have nonnegative spectral contributions.

For this study, data were both created computationally and measured experimentally for investigating the efficacy of the proposed constraint on Classical Least Squares. Experimental data have been collected using measurements of anions relevant for Hanford processing: nitrate, nitrite, and sulfate at concentrations representative of the Hanford site. Additional anions unknown to the model calibration are included in the testing experiments: carbonate, oxalate, acetate, and phosphate. The goal of the constrained Classical Least Squares preprocessing is to improve quantification of the calibrated species in the presence of multicomponent mixtures.

Constrained Classical Least Squares is compared to other preprocessing methods in this work: Principal Component Analysis, an artificial neural network autoencoder, Blind Source Separation methods, and Spectra Residual Augmented Classical Least Squares [3]. It is shown that constrained Classical Least Squares outperforms these other methods when there are unknown species present, while also only requiring a single spectrum to operate.

This work has the potential to enable robust real-time monitoring of multicomponent streams like those present at the Hanford site. Nuclear waste processing remains an ongoing environmental, engineering, and political challenge. The results discussed in this work may be a step towards a faster and safer disposal timeline for the Hanford Site.

[1] I. A. Udugama et al., “The Role of Big Data in Industrial (Bio)chemical Process Operations,” Ind. Eng. Chem. Res., vol. 59, no. 34, pp. 15283–15297, 2020, doi: 10.1021/acs.iecr.0c01872.

[2] R. L. McCreery, Raman Spectroscopy for Chemical Analysis, vol. 157. 2000.

[3] D. M. Haaland and D. K. Melgaard, “New augmented classical least squares methods for improved quantitative spectral analyses,” Vib. Spectrosc., vol. 29, no. 1–2, pp. 171–175, 2002, doi: 10.1016/S0924-2031(01)00199-0.