I finally succeeded in tweaking JMRI Operation until I got what I wanted. Until now, I got LCL cars playing ping pong all over the layout and unable to go back to staging. As expected, it quickly clogged layout spots. The answer what rather simple. Since I wanted to receive LCL but also send LCL, I originally created a single commodity called "LCL". Sounds logical... Many spurs were scheduled to receive and ship such commodity. But as expected, when a car was loaded it was routed to the nearest available spot, thus on the layout instead of the far away destination in the staging area. It was a disaster in the making, the endless waltz of cars shuffling around the place forever started.
To address this issue, I simply created a LCL-IN load for inbound traffic and LCL-OUT load for outbound stuff. The car swapping ended since going to staging was the most logical (and short) way to handle the loads and empties. A human brain is able to tell LCL is going in and going out, but that must be explained flat out to the software
Another problem emerged as a result of saturating the layout with incoming cars. This was due to the inbound traffic being set using the Walthers carfloat maximum capacity which is well over sixteen 40ft cars. That's a lot of traffic for a small crowded terminal and soon enough, the place overflowed with cars. JMRI - even if it was against it's own scheduling rules - had to route cars back to staging. Since the staging spurs weren't designed to accept such loads, the cars stayed stranded forever on the carfloat (acting as a yard) searching forever for a spot that didn't even exist and slowly but surely reducing the carfloat capacity to bring new cars in and get others out. Just like a Tetris game going wrong, the layout was destined to be game over in a matter of time.
However, the solution was once again very simple. I found out Harlem Station was served - most of the time - by platform carfloats. These special carfloats have a platform in place of the central track. This implement is generally used to unload cars directly on the platform which is linked by a sliding ramp to a pier or a warehouse (look at "Pier Platform" in the glossary). This is particularly useful when cars are bound to a non rail-served marine terminal. Why Erie did use the platform carfloats to serve Harlem Station is a mystery, but I guess they had them and it was handy at that time.
But that type of carfloat plays to my advantage: the number of incoming car is reduced to 12 cars maximum. Given most pictures and statistics show an average of 6 to 8 car was typical, I changed JMRI so the longest train is made out of 12 cars. Also, I made sure to set a random value in such a way the number of incoming cars varies and is generally around 8 cars.
Finally, I better assigned customers' and storage spurs following what I could understand from old pictures and historic sources. It certainly isn't 100% accurate, but it is a fair enough approximation of the real thing. Moreover, with less but better routed cars, I got rid of the loaded cars stranded on the carfloat. Now, every car is routed correctly to its scheduled destination and goes back to staging as it should be.
Given this is my first serious effort using JMRI, I'm quite happy to have reached a decent level of proficiency in less than 4 days. Certainly, this is enough for a small switching layout, but I'm well aware I could go further with it, taking in account the days of the week and cargo-specific conditions. This will be done in due time thought I already made some adjustments that brings life and unexpected events.
On another hand, JMRI made me understand how much a small car fleet (about 60 cars) can get repetitive with a 55 cars layout capacity. As I said in my previous post, this is getting quite boring fast and it disrupts completely the suspension of disbelief. For this reason, I tracked down parts and decals to complete or kitbash cars in my collection that had no purpose. I'll probably be able to add about 11 new reefers and I still have decals to build more Erie boxcars. In itself it doesn't bother me since bashing and improving rolling stock is always a nice way to spend time in a creative way.
Until then, the next step is to clean out the rails and wheels for flawless operation and experiment first hand with the new switchlist. Jérôme is already working out the mechanical aspect of the fleet including trucks, wheels and coupler height. It seems I did a poor job protecting the wheelsets from over spray and it quickly gunked the rail leading to poor electrical pickup. Lesson learned!