• microgrids
• multi-year
• optimization
• stochastic

Governments of developing countries are struggling to foster universal electricity access in their territory. Past policies have prioritized the peri-urban areas where the high-density population justifies the costs of extending the national grid, but rarely the same approach is affordable for rural areas, where people live in poor conditions with low demand related to basic needs, like lighting and mobile-phone charging. Within this context, isolated power grids, so called mini-grids, based on renewable sources and limited amount of fuel have demonstrated to provide good quality electricity at a reasonable price, thereby enabling the socio-economic development of the local community. However, the high business risks and payback uncertainties still curb many private and public initiatives. This research aims at de-risking mini-grid projects and reducing the financial burden through the development of advanced planning, sizing and operating methods for rural mini-grids that minimize operating and investment costs, while addressing the stochasticity of load, renewable sources and fuel procurement through Monte Carlo simulations. A novel stochastic operating approach based on Monte Carlo scenarios was proposed, as well as a new classification of stochastic approaches reviewed in literature. Moreover, the effect of using predictive strategies in the design phase has been studied and the results suggested the possible use of simplified strategies to optimally design mini-grids. Finally, a multi-year stochastic dynamic technique is proposed to plan the design of a mini-grid under load growth, which is very steep and unpredictable in newly electrified communities. The methodology enabled deferring the installation of components as the demand grows, thus achieving important savings in term of net present costs over traditional techniques, even more than 35%.