Olawale Makanjuola1,2,Raed Hashaikeh2
New York University1,New York University Abu Dhabi2
Olawale Makanjuola1,2,Raed Hashaikeh2
New York University1,New York University Abu Dhabi2
The commercial deployment of membrane distillation (MD) as an alternative to reverse osmosis (RO) and multistage flash (MSF) or multi-effect distillation (MED) for desalination has been largely hindered by the high specific energy consumption (SEC) in MD systems. Many studies have succeeded in reducing the SEC in MD to a threshold value of around 1,000 kWh/m<sup>3</sup> by using energy recovery devices (ERD) and other techniques. While this is a significant improvement for MD, RO and thermal desalination present much lower SEC values (7 – 150 kWh/m<sup>3</sup>), and the ERD incorporated in MD systems will increase the complexity, equipment cost, and required real estate for MD operations.<br/><br/>Our previous works have indicated that using thermoelectric devices as a direct thermal source inside the MD module can reduce the SEC by more than 35% below the 1,000 kWh/m<sup>3</sup> threshold (i.e. < 650 kWh/m<sup>3</sup>) whilst greatly simplifying the MD system by eliminating the support subsystems generally required for heating and cooling in MD. The thermoelectric device doubles as both a direct thermal source to drive the process and as an ERD reducing heat exchange losses to near zero. By combining high-fidelity computational fluid dynamics simulation and mathematical modeling of the thermoelectric device with experimental validation, we developed a fairly good model for predicting the performance of a thermoelectric MD system.<br/><br/>We are currently investigating the possibility of embedding column-type p and n thermoelectric pellets inside the MD membrane itself. This thermoelectric membrane is expected to provide a more localized thermal source and sink for driving the MD process. This in turn is likely to lower the SEC when used in an MD system. A theoretical model developed from the constitutive relations under isotropic conditions while assuming steady-state conditions, one-dimensional heat flow, and constant properties for both the membrane and thermoelectric material allows us to predict the SEC (in kWh/m<sup>3</sup>) in an MD process driven by a thermoelectric membrane. The result indicates that while a thermoelectric membrane can potentially reduce the SEC by over 90% below the 1000 kWh/m<sup>3</sup> threshold for conventional MD equipped with ERD, the SEC cannot go below around 15 kWh/ m<sup>3 </sup>even for the best permeable thermoelectric membrane.<br/>Our current efforts are geared towards realizing a physical thermoelectric membrane. Several avenues are being explored to successfully embed thermoelectric bismuth telluride pellets inside electrospun membrane. Concurrently, we are also working to improve the current theoretical model by accounting for all pertinent thermal and electrical resistances. Contact resistance and small-scale effects are some of the phenomena that the improved model will account for.