Westinghouse Electric (Sweden)

A unifying and accurate model to predict critical heat flux (CHF) over a wide range of conditions has been elusive since wall boiling research emerged. With the release of the experimental data utilized in the development of the 2006 Groeneveld CHF lookup table (LUT), by far the most extensive public CHF database available to date (nearly 25 000 data points), the development of data-driven prediction models over a large parameter space in simple geometry (vertical, uniformly heated round tubes) can be achieved. Furthermore, the popularization of machine learning techniques to solve regression problems has led to more advanced tools for analyzing large and complex databases. This work compares three machine learning algorithms to predict the CHF database. For each selected regression algorithm (ν-Support vector, Gaussian process, and neural network), a set of optimized hyperparameters is applied. Among the investigated algorithms, the neural network emerges as the most effective, achieving a CHF predicted/measured factor standard deviation of 12.3%, three times less than that of the LUT. In comparison, the Gaussian process regression and the ν-support vector regression achieve a standard deviation of 17.7%, about two times lower than the LUT. Hence, all considered algorithms significantly outperform the LUT prediction performance. The neural network model and training methodology are designed to prevent overfitting, which is confirmed by data analyses of the CHF predictions. Finally, future development directions (including data coverage, transfer learning, and uncertainty quantification) are discussed.

Nuclear industry workers exposed to uranium aerosols may risk kidney damage and radiation-induced cancer. This warrants the need for well-established dose and risk assessments, which can be greatly improved by using material-specific absorption parameters in the ICRP Human Respiratory Tract Model. The present study focuses on the evaluation of the slow dissolution rate (ss, d−1), a parameter that is difficult to quantify with in vitro dissolution studies, especially for more insoluble uranium compounds. A long-term follow-up of urinary excretion after the cessation of chronic inhalation exposure can provide a better estimate of the slow-rate dissolution. In this study, two workers, previously working for >20 y at a nuclear fuel fabrication plant, provided urine samples regularly for up to 6 y. One individual had worked at the pelletizing workshop with the known presence of uranium dioxide (UO2) and triuranium octoxide (U3O8). The second individual worked at the conversion workshop where multiple compounds, including uranium hexafluoride (UF6), uranium dioxide (UO2), ammonium uranyl carbonate, and AUC [UO2CO3·2(NH4)2CO3], are present. Data on uranium concentration in urine during working years were also available for both workers. The daily excretion of uranium by urine was characterized by applying non-linear least square regression fitting to the urinary data. Material-specific parameters, such as the activity median aerodynamic diameter (AMAD), the respiratory tract absorption parameters, rapid fraction (fr,), rapid dissolution rate (sr, d−1), and slow dissolution rate (ss, d−1) and alimentary tract transfer factor (fA) acquired from previous work along with default absorption types, were applied to urine data, and the goodness of fit was evaluated. Thereafter intake estimates and dose calculations were performed. For the ex-pelletizing worker, a one-compartment model with a clearance half-time of 662 ± 100 d (ss = 0.0010 d−1) best represented the urinary data. For the ex-conversion worker, a two-compartment model with a major [93% of the initial urinary excretion (A0)] fast compartment with a clearance half-time of 1.3 ± 0.4 d (sr = 0.5 d−1) and a minor (7% of A0) slow compartment with a half-time of 394 ± 241 d (ss = 0.002 d−1) provided the best fit. The results from the data-fitting of urinary data to biokinetic models for the ex-conversion worker demonstrated that in vitro derived experimental parameters (AMAD = 20 μm, fr = 0.32, sr = 27 d−1, ss = 0.0008 d−1, f A = 0.005) from our previous work best represented the urinary data. This resulted in an estimated intake rate of 0.66 Bq d−1. The results from the data-fitting of urinary data to biokinetic models for the ex-pelletizing worker indicated that the experimental parameters (AMAD = 10 μm and 20 μm, fr = 0.008, sr = 12 d−1, fA = 0.00019) from our previous dissolution studies with the slow rate parameter step-wise optimized to urine-data (ss = 0.0008 d−1) gave the best fit. This resulted in an estimated intake rate of 5 Bq d−1. Experimental parameters derived from in vitro dissolution studies provided the best fit for the subject retired from work at the conversion workshop, where inhalation exposure to a mix of soluble (e.g., AUC, UF6) and relatively insoluble aerosol (e.g., UO2) can be assumed. For the subject retired from work at the pelletizing workshop, which involved exposure to relatively insoluble aerosols (UO2 and U3O8), a considerably higher ss than obtained in dissolution studies provided a better representation of the urinary data and was comparable to reported ss values for UO2 and U3O8 in other studies. This implies that in vitro dissolution studies of insoluble material can be uncertain. When evaluating the results from the retrospective fitting of urine data, it is evident that the urine samples acquired after cessation of exposure provide less fluctuation. Long-term follow-up of uranium excretion after cessation of exposure is a good alternative for determining absorption parameters and can be considered the most viable way for determining the slow rate for more insoluble material.

Corrosion and hydrogen pickup of zirconium alloys can be life-limiting factors for fuel rods in light water reactors (LWRs). Extensive work has been performed and is still in progress to improve the specification of cladding materials to enhance the performance as well as to further increase the understanding of the involved mechanisms and to model the corrosion and hydrogen pickup kinetics. In this scope, knowledge regarding the behaviors of quaternary alloys (Zr-Sn-Fe-Cr) with different chemical composition is relevant. Two of the experimental alloys studied here also include niobium. These R&D quaternary alloys, including Alloy A, Alloy B, and Alloy C, were corrosion-tested for a long duration in an autoclave at EDF's R&D laboratories. Tests were performed in a water-chemistry representative for the primary water in pressurized water reactors (PWRs) at 320°C, 340°C, and 360°C. The weight gains were measured periodically, and some specimens were removed for hydrogen pickup measurements. The effect of tin and niobium on corrosion is discussed with respect to other metallurgical parameters. Cyclic features of the corrosion kinetics are studied, and comparisons between transition kinetics and oxide stratification are discussed. The effects of tin and niobium contents on hydrogen pickup are also studied. For all materials, oxide thickness and hydrogen content are linearly correlated irrespective of the duration of the experiment. Finally, the results from the ultra-long-term autoclave tests are assessed using experience gained previously from irradiation of the same materials in commercial PWRs as well as in the Halden test reactor.