Mitigating check valve slamming and subsequentwater hammer events for PPFS using MOC

The method of characteristic (MOC) was adopted to analyze the check valve-induced water hammer behaviors for a Parallel Pumps Feedwater System (PPFS) during the alternate startup process. The motion of check valve disc was simulated using inertial valve model. Transient parameters including the pressure oscillation, local flow velocity and slamming of the check valve disc etc. have been obtained. The results showed that severe slamming between the valve disc and valve seat occurred during the alternate startup of parallel pumps. The induced maximum pressure vibration amplitude is up to 5.0 MPa. The scheme of appending a damping torque to slow down the check valve closing speed was also performed to mitigate of water hammer. It has been numerically approved to be an effective approach.     

Quantitative analysis modeling for the ChemCam spectral data based on laser-induced breakdown spectroscopy using convolutional neural network

Laser-induced breakdown spectroscopy (LIBS) has been applied to many fields for the quantitative analysis of diverse materials. Improving the prediction accuracy of LIBS regression models is still of great significance for the Mars exploration in the near future. In this study, we explored the quantitative analysis of LIBS for the one-dimensional ChemCam (an instrument containing a LIBS spectrometer and a Remote Micro-Imager) spectral data whose spectra are produced by the ChemCam team using LIBS under the Mars-like atmospheric conditions. We constructed a convolutional neural network (CNN) regression model with unified parameters for all oxides, which is efficient and concise. CNN that has the excellent capability of feature extraction can effectively overcome the chemical matrix effects that impede the prediction accuracy of regression models. Firstly, we explored the effects of four activation functions on the performance of the CNN model. The results show that the CNN model with the hyperbolic tangent (tanh) function outperforms the CNN models with the other activation functions (the rectified linear unit function, the linear function and the Sigmoid function). Secondly, we compared the performance among the CNN models using different optimization methods. The CNN model with the stochastic gradient descent optimization and the initial learning rate=0.0005 achieves satisfactory performance compared to the other CNN models. Finally, we compared the performance of the CNN model, the model based on support vector regression (SVR) and the model based on partial least square regression (PLSR). The results exhibit the CNN model is superior to the SVR model and the PLSR model for all oxides. Based on the above analysis, we conclude the CNN regression model can effectively improve the prediction accuracy of LIBS.     

A study of the characteristics of the acoustic emission signals for condition monitoring of check valves in nuclear power plants

The objective of this study is to demonstrate that a condition-monitoring system based on acoustic emission (AE) detection can provide timely detection of check valve degradation and service aging so that maintenance or replacement can be preformed prior to the loss of safety function. This research is focused on the investigation and understanding of the capability of the acoustic emission technique to provide diagnostic information on check valve failures.AE testing for a check valve under controlled flow loop conditions was performed to detect and valve degradation such as wear and leakage due to foreign object interference. It is clearly demonstrated that the distinction of different types of failure were successful by systematically analyzing the characteristics of various AE parameters.     

Development of a swing check valve model for a low velocity pipe flow prediction

A swing check valve is commonly used to prevent a reverse flow in the pipe lines of a nuclear power plant. The flow resistance by the swing check valve varies with the location of the swing disk in the velocity range lower than the required minimum velocity for a full opening of the swing disk, thereby the fluid flow is significantly affected by the dynamic motion of the swing disk. Such a phenomenon is very important to analyze safety issues, one of which is the gravity feed following a loss of the residual heat removal (RHR) which occurs during a mid-loop operation. This paper focused on the development of a new check valve model to enhance the capability of the thermal-hydraulic system code. A new angular momentum equation for the disk of a swing check valve is proposed. The proposed model is implemented into the MARS code and verified through a comparison of the simulation results with the experimental data. In particular, the results of the simulation for the gravity feed line are comparably consistent with the real test data performed in a nuclear power plant.     

BP神经网络算法预测多组分材料中子屏蔽效果方法研究

针对多组分中子屏蔽材料优化设计中蒙特卡罗模拟计算时间长而对算法效率的制约,讨论了利用BP神经网络算法快速预测材料中子屏蔽效果的方法。以复合材料300种随机质量组分和其对应的蒙特卡罗计算的剂量值组成训练样本,建立了典型的3层BP神经网络模型,其剂量预测值与样本值的绝对偏差在±2以内。对训练样本之外的验证样本,绝对偏差扩大到-6.4～5.2之间。偏差分布统计显示70%以上样本的相对偏差绝对值在2%以内,定性判断该神经网络模型的计算精度和泛化能力满足优化算法使用。使用交叉验证法对网络进行二次训练,可提高训练样本的计算精度,但扩大了验证样本的计算偏差,表明神经网络建立中还需要考虑样本的拟合程度和泛化能力的平衡。     

Analysis of CHF in saturated forced convective boiling on a heated surface with impinging jets using artificial neural network and genetic algorithm

In this paper, a three-layer Back Propagation (BP) algorithm artificial neural network (ANN) for predicting critical heat flux (CHF) in saturated forced convective boiling on a heated surface with impinging jets was trained successfully with a root mean square (RMS) error of 17.39%. The input parameters of the ANN are liquid-to-vapor density ratio, ρl/ρv, the ratio of characteristic dimension of the heated surface to the diameter of the impinging jet, L/d, reciprocal of the Weber number, 2σ/ρ_lu²(L − d), and the number of impinging jets, N_j. The output is dimensionless heat flux, qco/ρvHfgu. Based on the trained ANN, the influence of principal parameters on CHF has been analyzed as follows. CHF increases with an increase in jet velocity and decreases with an increase in L/d and N_j. CHF increases with an increase in pressure at first and then decreases. Besides, a new correlation was generalized using genetic algorithm (GA) as a comparison with ANN to confirm the advantage of ANN.     

Design and test of hydraulic vacuum breaker check valve for simplified boiling water reactor

In the current design of the simplified boiling water reactor, the vacuum breaker check valve is an important safety component. The vacuum breaker check valve is the only key safety components which is not passive in nature. Failure of this mechanical valve drastically reduces the passive containment cooling system cooling capability and hence containment pressure may exceed the design pressure. To eliminate this problem novel vacuum breaker check valve was developed to replace the mechanical valve. This new design is based on a passive hydraulic head, which is fail-safe and is truly passive in operation. Moreover this new design needs only one additional tank and one set of piping each to the wetwell and drywell. This system is simple in design and hence is easy to maintain and to qualify for operation. The passive vacuum breaker check valve performance was first evaluated using RELAP5. Then the passive vacuum breaker check valve was constructed and implemented in the PUMA integral test facility. Its performance was studied in a large break loss of coolant accident simulation test performed in PUMA facility.     

利用脉冲快热中子分析方法和神经网络识别爆炸物

研究利用PFTNA方法和人工神经网络来识别爆炸物的方法。该方法利用PFTNA来获取被检测样品的热中子俘获γ谱和γ快中子非弹性散射谱,然后通过已知能谱对人工神经网络进行训练后识别未知样品。利用这种方法,能够快速的对未知样品进行定性分析。     

Study on the characteristic points of boiling curve by using wavelet analysis and genetic neural network

Local singularity of a signal includes a lot of important information. Wavelet transform can overcome the shortages of Fourier analysis, i.e., the weak localization in the local time- and frequency-domains. It has the capacity to detect the characteristic points of boiling curves. Based on the wavelet analysis theory of signal singularity detection, Critical Heat Flux (CHF) and Minimum Film Boiling Starting Point (q_min) of boiling curves can be detected by using the wavelet modulus maxima detection. Moreover, a genetic neural network (GNN) model for predicting CHF is set up in this paper. The database used in the analysis is from the 1960s, including 2365 data points which cover a range of pressure (P), from 100 to 1000 kPa, mass flow rate (G) from 40 to 500 kg m⁻² s⁻¹, inlet sub-cooling (ΔT_sub) from 0 to 35 K, wall superheat (ΔT_sat) from 10 to 500 K and heat flux (Q) from 20 to 8000 kW m⁻². GNN mode has some advantages of its global optimal searching, quick convergence speed and solving non-linear problem. The methods of establishing the model and training of GNN are discussed particularly. The characteristic point predictions of boiling curve are investigated in detail by GNN. The results predicted by GNN have a good agreement with experimental data. At last, the main parametric trends of the CHF are analyzed by applying GNN. Simulation and analysis results show that the network model can effectively predict CHF.     

LONSA as a tool for loading pattern optimization for VVER-1000 using synergy of a neural network and simulated annealing

This paper presents a new method for loading pattern optimization in VVER-1000 reactor core. Because of the immensity of search space in fuel management optimization problems, finding the optimum solution requires a huge amount of calculations in the classical method, while neural network models, with massively parallel structures, accompanied by simulated annealing method are powerful enough to find the best solution in a reasonable time. Hopfield neural network operates as a local minimum searching algorithm; and for improving the obtained result from neural network, simulated annealing is used. Simulated annealing, because of its stochastic nature, can provide for the escape of the result of Hopfield neural network from a local minimum and guide it to the global minimum. In this study, minimization of radial power peaking factor inside the reactor core of Bushehr NPP is considered as the objective. The result is the optimum configuration, which is in agreement with the pattern proposed by the designer.     

Optimal artificial neural network architecture selection for performance prediction of compact heat exchanger with the EBaLM-OTR technique

Artificial Neural Networks (ANN) have been used in the past to predict the performance of printed circuit heat exchangers (PCHE) with satisfactory accuracy. Typically published literature has focused on optimizing ANN using a training dataset to train the network and a testing dataset to evaluate it. Although this may produce outputs that agree with experimental results, there is a risk of over-training or over-learning the network rather than generalizing it, which should be the ultimate goal. An over-trained network is able to produce good results with the training dataset but fails when new datasets with subtle changes are introduced. In this paper we present EBaLM-OTR (error back propagation and Levenberg-Marquardt algorithms for over training resilience) technique, which is based on a previously discussed method of selecting neural network architecture that uses a separate validation set to evaluate different network architectures based on mean square error (MSE), and standard deviation of MSE. The method uses k-fold cross validation. Therefore in order to select the optimal architecture for the problem, the dataset is divided into three parts which are used to train, validate and test each network architecture. Then each architecture is evaluated according to their generalization capability and capability to conform to original data. The method proved to be a comprehensive tool in identifying the weaknesses and advantages of different network architectures. The method also highlighted the fact that the architecture with the lowest training error is not always the most generalized and therefore not the optimal. Using the method the testing error achieved was in the order of magnitude of within 10⁻⁵-10⁻³. It was also show that the absolute error achieved by EBaLM-OTR was an order of magnitude better than the lowest error achieved by EBaLM-THP.     

Development of a supporting tool for optimal fuel management in research reactors using artificial neural networks

The present work investigates an appropriate way to solve the problem of optimizing fuel management in research reactors. A computer program has been developed, to automate this procedure. The program suggests an optimal core configuration, which satisfies to established safety constraints. The suggested solution was based upon the use of coupled methods in which a stochastic technique, simulated annealing, and an adaptive back-propagation neural network are connected to each other.An objective function was developed based on two performance parameters: cycle length, which can be determined through the evaluation of the effective multiplication factor K_eff and power peaking factor P_max. The system uses optimization of these two parameters to finds configurations in which K_eff is maximized, whether a penalty function is applied to limit the value of P_max which should be lower than the one given in the initial configuration.A series of tests have been performed using a modified core configuration of the benchmark 10 MW IAEA LEU (low enriched uranium) research reactor and the result achieved is the optimum configuration of the studied core.     

Predicting of slow noise and vibration spectra degradation in the IBR-2 pulsed neutron source using a neural network simulator

Employing a neural net model of the noise state of the IBR-2 reactor (JINR, Dubna, Russia) and a model of the vibration state of movable reflectors in the reactor we have predicted slow degradation in reactor noises. Operative diagnostics and prediction of the reactor noise behavior with time involves separating of the reflector degradation trend in power noises. We investigate two neural models. The first concerns the vibrations of the reflectors and the second is a simplified reactor noise model. The predicted results are close, in character, to the experimental data. They show that it is the movable reflectors that are mainly responsible for the degradation of power noises.     

Structures and elements for the decommissioning cost estimations of nuclear research reactors

Decommissioning cost estimation is a very important technique when designing and planning a nuclear facilities’ decommissioning project. Decommissioning cost estimation should be made according to the phases of the decommissioning activities and the installed components of the nuclear facilities.     

Development and testing of analytical models for the pebble bed type HTRs

The pebble bed type gas cooled high temperature reactor (HTR) appears to be a good candidate for the next generation nuclear reactor technology. These reactors have unique characteristics in terms of the randomness in geometry, and require special techniques to analyze their systems. This study includes activities concerning the testing of computational tools and the qualification of models. Indeed, it is essential that the validated analytical tools be available to the research community. From this viewpoint codes like MCNP, ORIGEN and RELAP5, which have been used in nuclear industry for many years, are selected to identify and develop new capabilities needed to support HTR analysis. The geometrical model of the full reactor is obtained by using lattice and universe facilities provided by MCNP. The coupled MCNP-ORIGEN code is used to estimate the burnup and the refuelling scheme. Results obtained from Monte Carlo analysis are interfaced with RELAP5 to analyze the thermal hydraulics and safety characteristics of the reactor. New models and methodologies are developed for several past and present experimental and prototypical facilities that were based on HTR pebble bed concepts. The calculated results are compared with available experimental data and theoretical evaluations showing very good agreement. The ultimate goal of the validation of the computer codes for pebble bed HTR applications is to acquire and reinforce the capability of these general purpose computer codes for performing HTR core design and optimization studies.     

X荧光谱与人工神经网络相结合对陶片产地识别的研究

本文采用X射线荧光分析技术测量考古陶片中的微量元素,利用样品的元素种类和含量的不同对考古样本的产地进行智能化识别.针对多元素共存、计数率低等谱分析困难,采用人工神经网络方法对所测X荧光谱进行学习和识别.样本总数为48片,来自8个省份、20个采集地.用两种不同的网络结构分别对两类地域划分的陶片学习和识别,对准确产地分类的样本,产地识别率为100%;对其余样本,识别率大于60%.本方法的产地识别结果是可行的.     

Development of automatic algorithm for localizing loose parts with a steam generator

In this paper, an automatic localization algorithm for estimating the impact location by loose parts in a Steam Generator using modified triangular method is proposed and applied to the impact test data of YongGwang Nuclear Power Plant Unit 3 Steam Generator. The algorithm, at first, was developed at the Mock-up system and modified to apply for the real plant. The Steam Generator is modeled as a cylinder shape and the modified method is used to find out the impact point of a loose part on the model. The result of estimated impact point applying to the developed algorithm has below about 5% average error. If the algorithm will be installed in the existing plant or next generation plant, the safety and reliability of Nuclear Power Plant will be improved.     

Prediction of the correct measured activity of 186Re and 188Re from reactor produced natural rhenium using an artificial neural network

To optimize the cost effectiveness of ¹⁸⁶Re and ¹⁸⁸Re production, which have recently been used as radio pharmaceuticals for therapeutic purposes, we designed an artificial neural network (ANN) to evaluate the activity of combined ¹⁸⁶Re + ¹⁸⁸Re. One of the production ways is the (n,γ) reaction of natural rhenium which leads to combined ¹⁸⁶Re + ¹⁸⁸Re. Using the counted activity of ¹⁸⁶Re + ¹⁸⁸Re mixtures by a well type isotope calibrator, the precise activity of ¹⁸⁶Re and ¹⁸⁸Re is obtained by the ANN. A back-propagation ANN was trained using 30 activities of mixed ¹⁸⁶Re + ¹⁸⁸Re. The performance of the ANN was tested by Early-Stopping validation method, and the ANN was optimized with respect to its architecture. The response of the ANN shows significant precision that may be used for medical application of ¹⁸⁶Re + ¹⁸⁸Re mixtures.     

Development of compartment models with Markov-chain processes for radionuclide transport in repository region

This paper presents a new radionuclide transport model for performance assessment and design of a geologic repository for high-level radioactive waste. The model uses compartmentalization of a model space and a Markov-chain process to describe the transport. The model space is divided into an array of compartments, among which a transition probability matrix describes radionuclide transport. While similar to the finite-difference method, it has several advantages such as flexibility to include various types of transport processes and reactions due to probabilistic interpretation, and higher-order accuracy resulting from direct formulation in a discrete-time frame.We demonstrated application of this model with a hypothetical repository in porous rock formation. First we calculated a three-dimensional steady-state heterogeneous groundwater flow field numerically by the finite-element method. The transition probability matrix was constructed based on the flow field and hydraulic dispersion coefficient. The present approach has been found to be effective in modeling radionuclide transport at a repository scale while taking into account the effects of change in hydraulic properties on the repository performance. Numerical exploration results indicate that engineered barrier configuration and material degradation have substantial effects on radionuclide release from the repository.     

Development of mathematical models and the dual-gauge principle for surface-type neutron moisture content gauges

Mathematical models are developed for the response of surface-type neutron moisture content gauges. Models are developed for both cadmium-covered and bare BF₃-filled proportional counter detectors so that the dual-gauge principle of measurement can be applied. This consists of the simultaneous solution of the two responses to minimize the effect of variations in sample composition.The response models are based on epithermal and thermal line and area flux models, where if flux as a function of radial distance r from the source is φ(r), then line flux is ∫φ(r) dr and area flux is ∫φ(r)r dr. The flux models are obtained from a solution of the three-group diffusion equation for a point source of fast neutrons at the surface of an infinite half-space and are expressed as inverse Hankel transforms. The models are checked by calculation and are verified with Monte Carlo and experimental results.To minimize the effects of variations in sample composition, represented by the weight fraction of thermal neutron absorber, two gauge response models are solved simultaneously for moisture and absorber weight fractions. This application of the dual-gauge principle is tested with experimentally obtained responses on samples of known density and composition. The results of this feasibility study are encouraging in terms of eventual commercial application of the response models and the dual-gauge principle.