That's not quite how I understood "50 days of empty battery", but fair enough. This is not something that can't be solved, though. Even the 90% estimate is the worst case estimate, since there's actually a high likelihood of 3 TWh of storage being sufficient for 100% of renewables in any given German year.
> My approach was very very simple: if production is higher than consumption, store energy. Energy storage is limited, so when the stroage is full, energy is lost. If consumption is higher then production, take the difference from the battery.
That's Sinn's approach ("waste no energy and store everything, no matter the cost"), which yields significantly higher total costs (and storage capacities) than Zerrahn's approach which actually takes costs into consideration and leads to much smaller storage capacities on the basis of throwing energy away or using it opportunistically being actually cheaper than storing it for later grid use.
> Also keep in mind that this ignores transport an heat.
Transport and heat actually simplify things significantly for the grid since charging of EVs in Europe can be shifted around by as much as several days while heat will have been largely solved by the EU's 2010 and 2012 energy efficiency directives.