Prediction of critical heat flux of saturated upward flow boiling water in vertical narrow rectangular channels

Abstract

A study, on the methods to predict the critical heat flux (CHF) of saturated upward flow boiling water in vertical narrow rectangular channels, has been conducted. For the assessment of various CHF prediction methods, 608 experimental data were selected from the previous researches, in which the heated sections were uniformly heated from both wide surfaces under the high pressure condition over 41 bar. By using ACE algorithm, the recent statistical analysis tool, new empirical correlations (New Correla-tion-A and B) were proposed. The CHF data are well predicted with standard deviation (SD) with 13.1% and root mean square error (RMSE) with 13.1%, which are more advanced, compared with those of Katto correla-tion. Through the ACE transformation, it is found that the effect of the channel shape parameters on the CHF is greater than that of others. Taking account of proper correction factors into AECL’s Look-Up Table (LUT), CHFs are calculated with SD of 11.8 % and RMSE of 11.9 %, which are a little more improved than those of New Correlation-A and B. The comparative study, with experimental data at the pressure condition of about 41 bar, shows that LUT is potentially applicable to low pressure condition and the pragmatic prediction method. To theoretically predict the CHF of saturated flow boiling water in circular channels, 6,058 experi-mental data, for which liquid film dryout (LFD) model has its own engineering effectiveness, were selected from KAIST CHF data bank. Review on the representative LFD models (Whalley, Levy, Wurtz and Mishima models) shows that it’s reasonable to define the initial condition of quality & entrainment fraction at onset of annular flow (OAF) as the transition to annular flow regime & the equilibrium value respectively, and the pre-diction error of LFD model is dependent on the accuracy of the constitutive equations of droplet deposition & entrainment. In the Katto model, the CHF data are predicted with SD of 14.0% and RMSE of 14.1%, but some data could not be calculated. Meanwhile, in the present LFD model, which is based on the constitutive equations developed by Okawa et al., the whole data are calculated with SD of 17.1% and RMSE of 17.3%. So, the present model was finally selected as the best LFD model to predict the CHF for narrow rectangular channels. For the assessment of the present LFD model for narrow rectangular channels, 284 data were selected similarly with the case of circular channels. By using the present LFD model, the CHF data are predicted with RMSE of 22.9% with the dryout criterion of zero-liquid film flow, but RMSE of 18.7% with rivulet formation model. This shows that the prediction error for narrow rectangular channels is similar with that for circular channels in the present LFD model.