Imagine if you could tell your car that it only needed 30% charge during the week because you were only going back and forth to work, so the other 70% could be sold back to the grid for a profit. The car and the smart grid automatically negotiate. On Friday the car makes sure it gets up to 100% so you can take that long weekend drive.
Sell back to the grid? If I had an electric car, I wouldn't be inclined to wear out the batteries (which have a finite life in terms of number of charge cycles) doing this.I once did a back of the envelope calculation using typical lifespan data for Li-ion batteries - I'd have to sell back at a ridiculously high price to justify the loss of battery life. And even practical storage schemes like pumped storage hydro need to sell at a higher price than they buy - a cost that typically isn't included in the cost of electricity estimates of wind and solar (which is fair enough as it's too dependent on other factors, but something that needs to be considered). Moot point though, as electric cars will stay uneconomic unless there's an huge and unexpected drop in battery costs or oil prices skyrocket.
We can only predict demand for large regions, not locally.
That's all we need, thanks to the grid averaging out demand fluctuations over space. The high reliability of the current grid is evidence for that.
Thing is that is never, every does die down for hours or days. At least, not everywhere.
Really? I've seen it happen on UK grid monitoring webpages (e.g. here)- wind generation under 10% of total installed capacity for hours at a time is quite common. In fact output varies quite savagely, presumably due to the cube-law dependence of output on wind speed. People often claim that "the wind is always blowing somewhere", but the data suggests "not enough it isn't".
With something like wind and individual turbine failing will drop maybe 20MW. Wind speed changes slowly, so if it is 20 knts now it won't be less than 18 or 19 knts in 20 minutes time, giving plenty of opportunity to spool up other sources.
Which other sources though? Renewables (except hydro and biomass, which are limited) can't be spooled up, they either generate or not depending on the availability of the resource. You can predict that you won't have power, but then you need to do something about it. Turning fridges off will only help on a much shorter timescale.
In contrast, handling a power station failure is a solved problem with the existing grid, as are demand surges. Solving problems that have already been solved isn't a compelling reason for something.