Comparison of Adsorption-based Desalination Plant Performance Models

Jun W. Wu, Mark J. Biggs, Eric J. Hu

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Adsorption-based desalination (AD) is attracting increasing attention because of its ability to use low-grade thermal energy to co-generate fresh water and cooling. Various levels of model ranging from the simplest thermodynamic model through to the most sophisticated three-dimensional multiphysics models can be deployed to predict the performance of AD plants. Whilst the simplest of these models are unlikely to predict the performance as reliably as the more complex, their low computational cost means they are ideal for scoping analysis. Comparison of these simpler models against experiment and more complex models is, therefore, warranted. In this paper, we compare a thermodynamic model developed by the authors with an experimentally-validated lumped (i.e. zero-dimensional, or 0D) kinetic model developed by Thu et al. (2009). The predicted values of the specific daily water production (SDWP) and performance ratio (PR) at various heat source temperatures of a two-bed silica gel AD system from two models are compared. Good agreement between the two models is found for the SDWP when the heat source (i.e. hot water inlet temperature) is lower and the operation cycle time is longer. It was found, however, that the thermodynamic model predicts more than double the SDWP of the 0D kinetic model at the shortest cycle times investigated here. Further comparison indicates that whilst the thermodynamic model over-predicts the PR, the differences are modest.
Original languageEnglish
Title of host publicationChemeca 2010: Engineering at the Edge; 26-29 September 2010, Hilton Adelaide, South Australia
Place of PublicationBarton, A.C.T.
PublisherEngineers Australia
Number of pages9
ISBN (Print)9780858259713
Publication statusPublished - 2010


Dive into the research topics of 'Comparison of Adsorption-based Desalination Plant Performance Models'. Together they form a unique fingerprint.

Cite this