First what is your definition of full utilization and what aspect of the system are you talking about being fully utilized.
The Dual Contract specification was 40 tph operation or 90 second headways. This 90 second headway figure (including a 30 second allowance for station dwell times) was reaffirmed by the MTA in the Manhattan East Side Transit Alternatives MIS/DEIS.
This capacity figure is derived by using acceleration and deceleration rates, station dwell time and safety margin between trains. Most systems are pretty much the same on these aspects. Therefore it is not surprising that London Transport would state: "The services in the central area during peak periods are usually based on regular intervals ranging from 2 minutes down to 1½ minutes, the latter figure being in practice minimum interval which can be continuously worked on a single track..." Similarly the Moscow subway operates with 90 second headways during rush hours.
The capacity of a line is limited by its terminal facilities because it usually takes more than 90 seconds to turn around a train at most terminals. The maximum capacity for a stub terminal is 30 tph, assuming it takes 60 seconds to clear the crossover. It's 26 tph, if it takes 90 seconds to clear the crossover. The best that I know for NYC is 25 tph. A terminal with tail tracks sufficiently decreases the time over the crossover, so that it can handle 40 tph. Loop terminals also do not limit line capacity.
Second unequal headways on crowded lines cause problems, althought I will never prove that to your satisfaction.
You misunderstood my previous posts. I've shown theoretical basis for the assertion that uniform headways are essential. I've also pointed out inexpensive apparatus, used on other systems, to help operating personnel adhere to schedule.
Third look at a busy highway, when there is an entrance without adding a lane traffic slows down.
This should not be a valid comparison because a railroad has control over what is at the merge point and when. This is not true of a highway which usually cannot control arrival times at a merge point. If we assume that it still takes 1 minute for a train to clear a mergpe point, we can still have 40 tph operation with 20 tph on each branch. Simply schedule each branch to have 3 minute headways and also schedule each branch with a 90 second offset at the merge point. However, if the schedule makers schedule schedule the two trains to be at the merge point at the same time (like the TA has at the 11th St Tower), then this degenerates to your highway analogy.
Fourth how do YOU know otherwise. You load your question by assuming 75%. You dismiss all my methodology without saying how it is wrong. You accept no questions or critiscisms of you methodology.
Let's re-examine your assertion that changing the W from express to local will help and mine that it won't. Let's assume that all the services, N and W are operating with 8½ minute headways (7 tph) and the R is running with 6 minute headways (10 tph). That's fairly close to reality - I've added an extra R to make the numbers come out simpler.
Let's look at the W as a local scenario. You want to have equal headways on the Astoria line. So you will alternate the N and W service at 3 minute intervals. Trains will leave at: 8:00:00(N), 8:04:15(W), 8:08:30(N), 8:12:45(W), etc. The trains will arrive at Queensboro Plaza at these times plus a fixed offset. When should the R arrive? The earliest an R can arrive after the first N without being delayed (40 tph) is 8:01:30(R). The following R's will arrive at: 8:07:30 - 6 minutes later The following R will arrive at 8:13:30. The third R will be delayed by the W that arrived at 8:12:45 which will not create a clear path for its follower for 90 seconds or until 8:14:15. You can try any offset for the R you choose; there will always be a merging conflict.
Let's look at the W as an express scenario that I suggested with 1:1:1 3-way merging. This means that each line operates at 8 tph or 7½ minute headways. The optimum separation in the 60th St Tunnel is: 2½ minute headways (60min/24tph). Let the first R arrive at 8:00:00. Succeeding R's will arrive at 8:07:30, 8:15:00, 8:22:30, etc. Let the W's arrive at: 8:02:30, 8:10:00, 8:17:30, etc. Let the N's arrive at: 8:05:00, 8:12:30, 8:20:00, etc. There are no merging conflicts, as one would expect for equal merging. What about the headways on the Astoria Line? At Queensboro Plaza: 8:02:30(W), 8:05:00(N), 8:10:00(W), 8:12:30(N), 8:17:30(W), 8:20:00(N), etc. There is a 2:1 imbalance at one station. The W's will get more customers at QBP because its headway is 5 minuts as opposed to 2½ for the N. This is not bad because the W is less crowded because it has made only 2 previous stops. What about Ditmars and Astoria Blvd? There is a 1 minute difference in running time between the local and express. So leaving Ditmars and Astoria Blvd the times will be: 8:03:30(W), 8:05:00(N), 8:11:00(W), 8:12:30(N), 8:18:30(W), 8:20:00(N). Again, more people will use the W because of the unequal headways on different tracks. Again, this is exactly the desired behavior.
You evidence is 40 YO timetables which were from the days of the ABD.
I remember how things were 40 years ago. They were better. It's not a faulty memory - the data show it.
You make demands for service that need alomst twice the level of staffing with no way to pay for it.
The same number of trains and manpower were required for both scenarios. I added 1 extra train for numerical simplicity only. If the time bracket that David had chosen were 7:59 - 8:59, instead of 8:00-9:00, such an adjustment would not have been necessary.