An unofficial blog about Caltrain and California HSR talks about one solution to the capacity problem and some non-solutions. I personally take it with a grain of salt, but it's still interesting: http://caltrain-hsr.blogspot.com/2013/02/the-virtues-of-widt...
I hear they're trying to move all the stations to grade separated ones though, which should alleviate that.
Also, you'd have to get the public accustomed to "want to get out here, have to be in these cars", which, knowing the public, will take a big information campaign and still end up with people frantically charging through cars and screaming at conductors to open the doors...
There are 4 (edit: looks like 5) unused Metrolink cars sitting in the siding at 4th and King. I don't know why they aren't being used on some of the busier routes.
there is likely a point where the decrease in runs_per_hour is offset by the increase in capacity_per_run.
This assumes that the users will still use the service as long as the time_per_run (how long it takes to get from point A to point B) doesn't increase to a point where users choose to travel by different means.
For example, currently caltrain consists are 1 locomotive + 5 cars. Locomotives range from 2,500 usable horsepower (after HEP) to 3,600.
The highest horsepower locomotives, the MP36PH-3C's, are pretty much reserved for the baby bullet service. This means that to add an additional car to a train would increase weight on average by 20% - a big addition to a train that has is running flat out at nearly all times.
The problem that caltrain faces is pretty simple: they can't expand capacity without significant investment of capital into additional locomotives and rolling stock.