Is there a limit to the number of moving consists?

The frustrating thing about all of this is that it so inconsistent, therefore heard to explain unless you have experienced it. For instance i have two trains with exactly the same instruction set on the same route.the second train follows the first train by three or four signals - maybe one base board in distance. The first train runs along fine along the entire route. The second train, using exactly the same instructions goe into spastic shock. It stops for no reason, it reverse direction and proceeds in the wrong direction for as long as you let. If i am lucky I can stop the train and the “continue” the schedule correctly until it closes in on the first train, and then it starts the random cycle of events all over. The real problem is if you are running or instance 30 AI controlled trains on a route you spend all your time straightening things out and no time enjoying route.

This is a difficult thing to report to quality control since the issue is not consistent from one route to another. I have also found that the problem is also somewhat dependent upon the length of the train. Shorten the length of the first train and sometimes the following trains will behave less erratic.

I have found that Autodrive totally eliminates the problem when trains are following one another, but it takes many hours of work to setup the track marks and switch junctions properly for it to be successful. I just hope the next update doesn’t destroy Autodrive.
That is indeed very frustrating. Creating a test route with only a couple of consists, usually from my experience doesn't exhibit the problem which means a complete route needs to be sent to the QA Team for testing.

Your set up is similar to the set up I have with my Gloucester Terminal Electric tram/trolley route. Using the Schedule Library, I assign the same schedule to many drivers. They follow each other at a distance and stop at the stations on the line. Depending upon the route, they either use a turn loop to go back in the opposite direction or use a stub end where they pull in, wait and crossover on exit to return in the opposite direction.

Prior to this update, the session worked fine. You summed it up perfectly.