Remote areas throughout the world often face great difficulties with regards to a cost-effective production of electricity and drinking water. To date, it is mostly done by means of diesel generators which are however subject to volatile market prices, high operational cost, logistical supply problems and environmental concerns. Stand-alone microgrids powered by renewable energy sources have the potential to overcome these drawbacks. In this context, SolteQ Energy has developed the FreshWaterMill system (FWM) which can provide an isolated community with both electricity and fresh water. It involves a wind turbine with a hydraulic transmission system for water desalination via the process of reverse osmosis (RO) in addition to a generator which is placed in parallel to the RO plant. The generator is seconded by a PV system and a battery bank for electricity production. The main constraints for the implementation of renewable energy systems are the high initial cost, a problem especially faced by developing countries. However, in contrast to developed countries, their system reliability requirements for power system design are less strict which allows for more freedom in finding an optimal compromise between system cost and reliability. This trade-off approach has the potential to substantially lower the overall cost as stand-alone renewable energy systems tend to be oversized due to the fluctuating and unpredictable nature of solar and wind and the lack of controllable energy sources.
This thesis aims at optimally sizing the bespoke FWM system for specific locations by searching for a best compromise point between the conflicting objectives, cost and system reliability. For this purpose, a model of the FWM system and a suitable operation strategy has been developed and the triple-objective optimisation problem was defined which simultaneously minimises the net present cost (NPC) as well as the loss of power and water supply probability. The optimal sizing procedure is performed by means of the Pareto-based multi-objective genetic algorithm (GA) solver ‘gamultiobj’ part of the Matlab Optimisation Toolbox. The Pareto-based multi-objective method provides the system designers with numerous FWM system configurations along with the optimal size of each component allowing them to find one or more appropriate solutions from a set of alternatives based on specific requirements of the location. GA has been chosen because compared to classical optimisation techniques it is highly applicable to complex and multi-modal problems with a large search space and is able to tackle these with little computational effort. The simulations of the FWM system are run for an entire year and require the input of hourly solar radiation, wind speed, ambient temperature, electricity and fresh water demand of the location under study as well as technical and economic specifications for each component. The simulation and optimisation procedure has been applied to a real case study of a remote Colombian island in the Caribbean Sea with a yearly average wind speed of 7 m/s.