Roberto Schirru

A new approach to the use of genetic algorithms to solve the pressurized water reactor's fuel management optimization problem

Abstract A Genetic Algorithm (GA) based system, coupling the computer codes GENESIS 5.0 and ANC through the interface ALGER has been developed aiming at pressurized water reactors (PWR) fuel management optimization. An innovative codification, the List Model (LM), has been incorporated into the system. LM avoids the use of heuristic crossover operators and only generates valid nonrepetitive loading patterns in the reactor core. The LM has been used to solve the Traveling Salesman Problem (TSP). The results got for a benchmark problem were very satisfactory, in terms of precision and computational costs. The GENESIS/ALGER/ANC system has been successfully tested in optimization studies for Angra 1 power plant reloads.

The Ant-Q algorithm applied to the nuclear reload problem

Abstract The nuclear core fuel reload optimization is a NP-complete combinatorial optimization problem where the aim is to find a pattern of fuel assemblies that maximizes burnup or minimizes the power peak factor. For decades this problem was solved using an experts knowledge. From the eighties, however, there have been efforts to automate fuel reload. The first relevant effort used simulated annealing, but more recent efforts have shown the genetic algorithms (GA) efficiency on this problem. Following this trend, our aim is to optimize nuclear fuel reload using Ant-Q, a reinforcement learning algorithm based on the Cellular Computing paradigm. Ant-Qs results on the traveling salesmen problem, which is conceptually similar to fuel reload, are better than the GAs. Ant-Q was tested on fuel reload by the simulation of the first out-in cycle reload of Biblis, a 193 assembly PWR and preliminary tests were performed for the cycle 7 reload of Angra I PWR. Comparing Ant-Qs results with the GAs, it can be verified that, even without local heuristics, the former algorithm can be used to solve the nuclear fuel reload problem.

Basic investigations related to genetic algorithms in core designs

Abstract The use of genetic algorithms as a tool to improve nuclear reactor core design optimization is proposed. A method based on genetic algorithms are developed to solve optimization problems that involve cell parameters adjustment. In this paper we address a global optimization approach to nuclear reactor core design problems. In previous work, a great variety of local optimization method have been applied in such a kind of problems. In order to illustrate the performance, we apply the method based on genetic algorithms to two traditional test problems that have been considered in literature. These test problems describe the principal framework of the approach proposed, because they have known optimal solutions. Due to the global exploitation capacity and independence of prior knowledge of the search space characteristics proportioned by the genetic algorithms the application scope was extended. The method was applied to optimize a more complex problem, to which traditional methods do not apply. The goal of this third problem is to minimize the average peak factor in a three enrichment zone reactor by the adjust of the fuel radii, cladding thickness, equivalent radii, the three zone enrichments, fuel material and cladding material. In this case a qualitative analysis of the results is made.

Prediction of volume fractions in three-phase flows using nuclear technique and artificial neural network

This work presents methodology based on nuclear technique and artificial neural network for volume fraction predictions in annular, stratified and homogeneous oil-water-gas regimes. Using principles of gamma-ray absorption and scattering together with an appropriate geometry, comprised of three detectors and a dual-energy gamma-ray source, it was possible to obtain data, which could be adequately correlated to the volume fractions of each phase by means of neural network. The MCNP-X code was used in order to provide the training data for the network.

The fuzzy clearing approach for a niching genetic algorithm applied to a nuclear reactor core design optimization problem

Abstract This article extends previous efforts on genetic algorithms (GAs) applied to a core design optimization problem. We introduce the application of a new Niching Genetic Algorithm (NGA) to this problem and compare its performance to these previous works. The optimization problem consists in adjusting several reactor cell parameters, such as dimensions, enrichment and materials, in order to minimize the average peak-factor in a three-enrichment zone reactor, considering restrictions on the average thermal flux, criticality and sub-moderation. After exhaustive experiments we observed that our new niching method performs better than the conventional GA due to a greater exploration of the search space.

Particle Swarm Optimization in Reactor Core Design

Abstract This work presents particle swarm optimization (PSO) as an alternative method for solving an optimization problem that arises during nuclear reactor core design. The method is introduced and applied to a simplified core optimization problem found in literature. When compared with other evolutionary computation–based methods, PSO performs better. Moreover, PSO presents easier modeling and demands less computational effort in the optimization process. The obtained results are shown and discussed.

Explosives detection using prompt-gamma neutron activation and neural networks

This work describes a study of the application of a neural network to determine the presence of explosives using the neutron capture prompt gamma-ray spectra of the substances as patterns which were simulated via Monte Carlo N-particle transport code, version 4B. After the training of the neural networks, it was possible to determine the presence of the C-4 explosive, even when they were occluded by several materials. The neural network was a powerful tool, able to recognize prompt gamma-ray explosive patterns in spite of the presence of occluding materials. Besides that, the network was able to generalize, identify the presence of explosive in cases in which it had not been trained. In that way, it was revealed as a potential tool for in situ inspection systems.

Genetic algorithms applied to turbine extraction optimization of a pressurized-water reactor

In this work, we propose the use of a genetic algorithm (GA) for the determination of the optimal fraction of mass flow rate to be extracted from each stage of the turbines of a typical pressurized-water reactor (PWR) secondary side, in order to increase cycle efficiency. Here, we show some preliminary results obtained in a case study in which the PEPSE® system was used as simulation tool.

Adaptive vector quantization optimized by genetic algorithm for real-time diagnosis through fuzzy sets

The accurate diagnosis of accidents in a nuclear power plant has fundamental importance for decision making necessary to mitigate their consequences for the power plant as well as for the general public, on the basis of emergency planning. Two main characteristics should be achieved in this kind of diagnostics, namely, real-time features and adaptive capacity. The first characteristic gives the operators the possibility of predicting degraded operations and monitoring critical safety functions related to that specific situation. The second one allows the system to be able to deal with accidents that were not incorporated in the training sample set, in which case the operators are unprepared because they were not trained to face an event that they did not observe even in simulator training. The Three Mile Island accident is a classic one to demonstrate that these kinds of events are possible. Several methodologies have been tried to match those characteristics. While the first one is achieved through the permanent evolution ofnew faster processors, the second one can only be achieved through the simulation of human cognitive processes, which show higher adaptive behavior. Our model utilizes a neural network, fuzzy sets, and a genetic algorithm to simulate that behavior. We have used a neural network activated by an additive model and trained with an unsupervised competitive training law. Once trained with three accidents (steam generator tube rupture, blackout, and loss-of-coolant accident), a synaptic matrix was obtained, in which the elements represent the interchanging weights between neurons in the concatenated input/output space and the competitive neurons that fight to encode the input-output vector. This kind of competition establishes a statistical classification of the state variables, changing with time, clustering them in centroids labeling the kind of accident for which variables are being sampled. Thus, the accident identification is done in real time with the synaptic matrix. However, the centroids are located in the same time value, in view of the fact that the neural network algorithm treats the variable time as an independent one. Therefore, a genetic algorithm is applied to a fuzzy system formed by the partition of the variables space with fuzzy sets determined by the neural network centroids, in order to estimate the optimal positions in the time variable where the fuzzy system centroids must be located. As a consequence, the diagnostic can be done in representative regions of each accident with maximum confidence.

PWR's xenon oscillation control through a fuzzy expert system automatically designed by means of genetic programming

This work proposes the use of genetic programming (GP) for automatic design of a fuzzy expert system aimed to provide the control of axial xenon oscillations in pressurized water reactors (PWRs). The control methodology is based on three axial offsets of xenon (AOx), iodine (AOi) and neutron flux (AOf), effectively used in former work. Simulations were made using a two-point xenon oscillation model, which employs the non-linear xenon and iodine balance equations and the one group, one-dimensional neutron diffusion equation, with non-linear power reactivity feedback, also proposed in the literature. Results have demonstrated the ability of the GP in finding a good fuzzy strategy, which can effectively control the axial xenon oscillations.