The World Health Organization predicts that by 2050 up to four billion people (nearly two-thirds of the world’s present population) will face fresh water shortages. Drinking water scarcity is already posing major problems for more than a billion people, mostly in arid developing countries. As climate change continues to redistribute global rainfallpatterns such conditions will become increasingly common.
Desalination of seawater has long been used in regions of water scarcity to supplementthe supply of clean water for human consumption and irrigation. The most common method is reverse osmosis: forcingseawater through semi-permeable membranes that filter out the dissolved salts and fine solids. The major factor limiting the efficiency of this process, and the rate of fresh water production, is the high pressure, and thus energy, required to drive water through the membrane itself. This hinders the large-scale adoption of seawater desalination for water-stressed populations.
However, an alternative to conventional membranes exists that could produce a major increase in the efficiency of the reverse osmosis process. Recent molecular dynamics simulations (see, e.g., ) indicate that dense arrays of carbon nanotubes (CNT) could be used as a nano-filtration membrane. The image shows work undertaken by Programme Researchers on simulating the flow of simple molecules through a section of CNT membrane. The CNT proves highly efficient at repelling ions and other contaminants (which face a large energy barrier at the inlet of the CNT), butpresentslittle impediment to water molecules . The simulations indicate that there is a massive decrease in the resistance to water flow as compared to that predicted by conventional theory. Such results suggest that nano-filtration membranescould transform the cost/benefit analysis forlarge-scale desalination and other water purification projects.
While molecular simulations can explore the basic principles, for the engineering design of a complete reverse-osmosis system the macroscopic flow characteristics must also be modelled: the fluid interactions with the CNTs at the molecular level will affect the macroscopic flow behaviour, and vice versa. In simulations in the programme, molecular dynamics applied near and within the membrane will run concurrently with a continuum treatment of the reverse-osmosis system on a macroscopic scale. Importantly, molecular dynamics will not be performed over the entire spatial domain; exploiting scale separation means that fluxes will only be computed via molecular simulation at points on the surface that are close enough to distinguish macroscopic flow variations.
 Corry B. 2008. Journal of Physical Chemistry B 112:1427-1434