This interesting article from MIT Technology Review series "Ten Breakthrough Technologies 2014" suggests that, using information technology to generate sophisticated predictive models of how much wind and solar will be generated in 15-minute increments, electricity grids will be able to support MUCH MORE "intermittent" renewable energy than previously thought.
The (US) National Center for Atmospheric Research (NCAR) in Boulder, Colorado has a pilot project, whereby Xcel, one of the US' largest utilities and the main power distributor in Colorado, is able to avoid the need for back-up fossil fuel generation, taking data from every wind turbine feeding into its system and using the NCAR models. Based on the new models, Xcel supports a mandate for utilities to get 30% of their electricity from intermittent, renewable sources.
And NCAR is now working on a similar model for solar, to match the wind models. The main challenge is the lack of data from residential rooftop systems.
And the MIT article mentions one of my favorite ideas -- the use of electric car battery storage to offset dips in the grid from use of renewables.
These types of developments are among the reasons why U.S. technology thinkers now see renewables as no longer expensive or impractical, and why the largest share of U.S. new generation added last year was renewable wind/solar.