More Red Signal Testing

[boleyd]

The route is MSGSAPPER's from Portland Maine north toward the Canadian border. Looks like a medium sizes route wiyh end-to-end runs in the vicinity of 30 to 45 minutes.Very Heavily modified.

I removed all signals. I started 4 trains from the "Rigby yard" on the south end. Started all at the same time. All four arrived at Augusta, the Capital of Maine, with no scratches. Observation showed that the first train to get to a junction caused any oncoming train to wait until the junction was cleared. A faster train maintained "social distance" from any train ahead. All for trains arrived at Augusta.

Signals my be required in some circumstance but in this particular setup there were none and all objectives were met. Yes, there were delays and I believe that proper scheduling, rather than a "shotgun start" would allow reasonable predictability within the structure of a schedule.

I accidentals left two signals, one on each of the dual track, and trains stopped behind the first stuck train. Other experiments did not show any relationship due to the type of track, engine, or signal, including semaphores. Most RED delays seem to be "wait and proceed" either a few seconds or 1.5 minutes. Without a way to manage this logic I will stick to a naked route.

 

[llebrez]

Not a solution. But long ago I realized that if you take all the signals out of a layout, AI would operate sometimes better. Am I crazy? Perhaps. I found that unless you place signals very carefully, and test everything (takes a long time), AI will foul at one time or another. AI is not as bad as it may seems on basic operation. Make it complex and stupidity sets upon your operations. Take this basic example: two trains are coming head on, but there is a passing loop, so one train stops so the other takes the side track and proceeds. The one that stops, if no signal, will almost enter the points but stop, and the other will pass through it taking the side track. If you set a signal let's say 50 or 100 mts away from the points, the signal will be red and detain the consist there while the other passes on the side. So signals are needed if you want a near realistic operation. AI is "smart" enough so that two trains following one to the other, will not collide if one is faster than the other, without signals. I am sure I am not telling you anything new here, but someone has to say something about it.

 

[MP242]

If you notice on real railroads, not each and every turnout is signaled, especially in yards.

Invisible signals (such as "Invisible Signal-norfolksouthern37 <KUID:45324:24010>", or "Signal Thingy") can be used in a lot of places.

Yellow directional markers can be used, pointing towards trains, to have them avoid collisions.

 

[boleyd]

I agree with a strategic, and sparse, signaling strategy. While a route works better, in my case, with no signals, Once I get a schedule working I will probably install some signals "for effect". Signals are a large element in just a visual sense, which is really at the core of the program. If it was not a visual experience Trainz would not be at the top of list for a full service simulator.

One other possible problem area is that this route had a lot of "X" full crossovers that included the four signals need to manage them. I removed them and installed single crossovers where they are really needed, and in a direction that is rational. This is done by using the industries and stations as a guide to provide ONLY the single crossovers to support those activities. A railroad is not a free-for-all. Companies only lay track, and signal, where income and safety dictate such expense. Looking at the route I adopted, considerable savings can be had once you realize that a "grain train" does not need to access a station. You also make either freight or passenger the prime objective when setting a schedule. There may be times that no trains are moving simply because there is no commercial reason for movement. Fewer signals and crossovers should help.

I should mention that the route is ALL dual track, which, while flexible, requires more management (signals and crossovers). I may go to single in some sections to "save money" and use sidings. Have to do a "cost analysis" to see if that is cheaper while supporting profit goals. Maybe some day I can ask what the cost is, on a ton per mile, to handle a grain contract. Especially when that would require a new siding so the customer's commitments are met in a cost effective manner.