Modeling of quality indicators for water demineralization process at power reactors

Abstract

Regression models of demineralizing filters for a nuclear power plant will increase the efficiency of the desalination process for turbine condensate. To identify the control object, experimental and statistical research methods were applied. To detect the presence and study the extent of relations between the process factors and output variables, the STATISTICA (StatSoft) analytical system was used. The impact of factors on the studied traits was assessed using a nonlinear estimation module. Upon the analysis of the desalination plant operation, the following initial parameters were identified. These parameters influence the control decisions in managing the object, can be predicted over time and are not random variables such as the values of specific conductivity and hydrogen index. For specific electrical conductivity: Xh is the specific electrical conductivity of the condensate after the desalination plant; Na is the concentration of sodium ions in the condensate behind the condensate electric pump of the first stage; CN2H4 is the concentration of hydrazine in the feedwater of the steam generator; CFe is the concentration of iron ions in the condensate on the condensate electric pump of the first stage; T is the temperature of the condensate sample on the condensate electric pump of the first stage; CCl is the concentration of chlorine ions in the purge water of the salt compartment of the steam generator; and CSO4 is the concentration of sulfate ions in the purge water of the salt compartment of the steam generator. For the hydrogen index: pH is the hydrogen index of the condensate after the desalination plant; CN2H4 is the concentration of hydrazine in the feedwater of the steam generator; T is the temperature of the condensate sample on the condensate electric pump of the first stage; and CFe is the concentration of iron ions in the condensate behind the condensate electric pump of the first stage. To detect the presence and study the extent of relations between the selected factors of relevant process and output variables, a correlation analysis was conducted and input factors were subsequently ranked. The correlation analysis allowed ranking all the variables and their possible interaction effects in a descending degree of impact on the outputs. In accordance with the selected system outputs, based on the ranking of input variables, regression analysis was performed and corresponding mathematical dependences were obtained with alternate inclusion of factors in a decreasing degree of their influence on the output variable to establish functional dependencies between the experimental data on desalination plant operation. The regression dependence for the experimental data of specific conductivity on the selected state parameters was obtained. The coefficient of determination is equal to 0.7245. The regression dependence describing the change of the hydrogen index has also been obtained. The coefficient of determination is equal to 0.7231. The proposed models explain 72% of the variation in the dependent variable. The correlation ratio is 0.82, which determines the close relationship between the values. Regression models adequately describe the operational data of the nuclear power plant and can be used as part of the control system of the filter unit.