Search

Condensation on banks of tubes is of considerable interest in the power, refrigeration and process industries, where large scale condensers form a significant proportion of plant capital costs. Since the pioneering paper by Nusselt in 1916, numerous investigations, both experimental and theoretical, have made great inroads into the understanding of the important physical factors effecting performance, including effects of vapour shear and condensate inundation on heat-transfer performance. Despite this there are still significant gaps in the knowledge and no single recognised design approach exists for condensers under all conditions. Purely theoretical models have shown some success in modelling condensation on single tubes under high shear regimes, but these have not been successfully extended to full tube banks. The present work begins by drawing together a comprehensive data base of experimental results from those available in the literature. This includes assessing the experimental accuracy of the data and organising it into a consistent format to allow detailed comparison with existing and future models. The resulting data base comprises 13 tube bank geometries, 7 test fluids and over 4000 individual data points. The data base was used to evaluate existing theoretical and empirical models, and highlighted the weaknesses therein. In particular, it was found that empirical approaches were limited to application (ie refrigeration or steam condensers), with development and validation being based on experimental data for single fluids or groups of fluids. When these models were compared to the more comprehensive data base described above their performance was poor. Consequently there is limited confidence in their extension to applications outside those they were developed for. A new empirical based model was then developed. The approach involved identifying relevant dimensionless groups to account for the various physical factors which may affect heat transfer during condensation on a bank of tubes and formulating these into an equation involving a number of initially unknown but empirically obtainable constants. An iterative scheme was then employed to eliminate those groups having little effect on the result while retaining those which proved to be more important. The resulting model predicted the majority of the experimental data base to around 12%. A subsequent parametric study 3 showed the correct dependence of heat transfer coefficient on factors such as vapour velocity and tube row. The thesis concludes with some suggestions for future work.