Mirko Skiborowski, Korbinian Krämer, Adel Mhamdi, Wolfgang Marquardt:
Structural optimization of hybrid RO-MED desalination plants
13. AMK (Aachener Membran Kolloquium), Aachen, 27-28.10.2010
Today there are already about three billion people having no access to potable water while over 97% of the worlds water resources are reposing in the oceans. Membrane based processes and thermal desalination plants are two alternatives to overcome the problem of fresh water supply by means of seawater desalination. Reverse Osmosis (RO) is the major membrane-based seawater desalination technology and accounts for approximately 56.5% of the worlds seawater desalination capacities . As a thermal desalination technology, multieffect distillation is stated to be more flexible and less energy consuming than the common multi-stage flash process . The performance and cost of a desalination plant depend on the type of plant, the interconnection of the different unit operations and their parameterization and site specific characteristics. Membrane manufacturers provide engineering tools to support the engineer in the design of stand-alone RO plants utilizing their membrane modules and software like WTCost© of the Bureau of Reclamation. For the cost estimation, however, only structural optimization allows to consider all impacts at once and find an economical optimal desalination plant . The optimal design of a hybrid desalination plant, as the combination of a reverse osmosis network (RON) and a forward feed multiple-effect distillation (FF-MED) plant is considered in this work. An extendable superstructure is combined with rigorous process models and a detailed techno-economical model of the whole plant. For the RON, pressure vessels containing a series of membrane modules, a pressure exchange system as well as pumps and turbines are considered as possible unit operations. A heuristic reduction of the superstructure and a novel initialization scheme are proposed to create a robust optimization model. The resulting mixed-integer nonlinear programming problem (MINLP) is then solved using the general algebraic modelling system (GAMS).
Desalination, Reverse Osmosis, Multiple-Effect Distillation, Hybrid Plant, Structural Optimization, MINLP