This paper goes into a bit more detail on the "pulse width modulation" scenario for controlling COVID-19 after the initial peak (among other things). It's...not especially cheery reading.

https://science.sciencemag.org/content/early/2020/04/14/science.abb5793 https://twitter.com/russss/status/1239680127061168141

To be absolutely clear - this is a "herd immunity" approach. The only other approach is to try and eliminate the virus entirely. And that is looking quite unlikely.

In pretty much the best case scenario, with double the current ICU capacity, and assuming seasonal forcing we could (probably?) reach that point with five periods of social distancing.

(These numbers are based on the US but other countries will be broadly similar)

Without seasonal forcing (reduction of transmission in the summer) we're looking at social distancing well into 2023.

This again assumes doubling of current ICU capacity.

These all assume a base wintertime R0=2.2, which I think might be a little optimistic.

The higher the R0, the higher the percentage of the population that needs to be immune in order to achieve herd immunity.

At R0 = 2.2 it's 54%
At R0 = 3.0 it's 67%
At R0 = 4.0 it's 75%

And this assumes that immunity is permanent, or at least long-lasting, which is by no means guaranteed.

They estimate the duration of immunity to two existing human coronaviruses (which cause the common cold) to be around 45 weeks.