| Reflux and vent condensers are vertical separators where film condensation occurs. A vapour mixture is supplied at the bottom of the tubes and encounters vertical cold surfaces. A falling film forms and exits from the bottom of the tubes, flowing counter-current to the vapour, but co-current to the coolant on the shell side. Vapour velocities at or above the critical cause flooding, that is, some or all the liquid exits from the top of the tubes rather than from the bottom. The calculation of this 'flooding velocity' is critical for the design procedure, since it imposes a limit on the condensation rate in a given condenser and constitutes the main disadvantage of these condensers. Most of the correlations to predict the flooding velocity come from isothermal experiments data and are not necessarily representative of vent condensation. The current design methods require minimum safety factors of 30% or more. The oversized condensers incur high capital and operational costs and could lead to unsatisfactory operation and even flooding at the design operational conditions. Experiments consisting of steam/air condensation at sub-atmospheric pressures were performed in a 3 m long, double-pipe vertical condenser with an internal diameter of 0.028 m. The experiments provide data of steady-state operation and flooding under condensation conditions. The former allows the evaluation of the design methodologies whilst the latter permits the evaluation of the flooding correlations. The correlation by English et al. is the most successful in predicting the flooding velocity in the vent condensation of steam/air. The deviation from the experimental flooding point varies with the vapour velocity. The results can be used to calculate more adequate safety factors, up to 50% lower than the currently recommended in the open literature. Two design methods are evaluated, namely the film and the equihbrium methods. For this purpose, a software tool was developed using VB.NET. After comparing the predictions of both methods, it becomes apparent that the film method should always be preferred when enough data are available. Previously unpublished recommendations for the computation of the methods are also reported. The deviation of the predicted heat-transfer area from the experimental one is shown to be a function of the Lewis number, where the best predictions are at Le = 0.68. At lower Lewis numbers both methods over-predict the area. This fact is commonly ignored when choosing a safety factor to enlarge the designed area, and in some cases it could be more appropriate to, in fact, decrease it. |
