PREDICTION OF 2-PHASE MIXTURE DENSITY USING ARTIFICIAL NEURAL NETWORKS

In nuclear power plants, the density of boiling mixtures has a signifi cant relevance due to its influence on the neutronic balance, the powe r distribution and the reactor dynamics. Since the determination of th e two-phase mixture density on a purely analytical basis is in fact im practical in many situations of interest, heuristic relationships have been developed based on the parameters describing the two-phase syste m. However, the best or even a good structure for the correlation cann ot be determined in advance, also considering that it is usually desir ed to represent the experimental data with the most compact equation. A possible alternative to empirical correlations is the use of artific ial neural networks, which allow one to model complex systems without requiring the explicit formulation of the relationships existing among the variables. In this work, the neural network methodology was appli ed to predict the density data of two-phase mixtures up-flowing in adi abatic channels under different experimental conditions. the trained n etwork predicts the density data with a root-mean-square error of 5.33 %, being similar to 93% of the data points predicted within 10%. When compared with those of two conventional well-proven correlations, i.e. the Zuber-Findlay and the CISE correlations, the neural network perfo rmances are significantly better. In spite of the good accuracy of the neural network predictions, the 'black-box' characteristic of the neu ral model does not allow an easy physical interpretation of the knowle dge integrated in the network weights. Therefore, the neural network m ethodology has the advantage of not requiring a formal correlation str ucture and of giving very accurate results, but at the expense of a lo ss of model transparency. (C) 1997 Elsevier Science Ltd.