Having noted how the last few releases from Trainsimulations (Mullan Pass, Rock Island) have reverted to MSTS physics instead of using Davis values (ORTS Selgman, Ruel, Milw) I wanted to see what the actual difference between the two sets of figures were demonstrating. A little bit of testing on flat level track shows how the Davis numbers were making the cars a bit too 'sticky' when stationary, requiring much more power to simply get them rolling.
I created two 90 car trains of 286k covered hoppers (about 12,500 US tons) and attached a single SD40-2. The brown BNSF cars are from Seligman 2 route and have Davis values. The identical gray CEFX cars are from Mullan route and have MSTS Friction values. The only editing done to wag files was to reduce the mass for the seligman car to match that of the mullan car.
With a solid train of Davis values, I could not get the train moving with the single SD40-2. Only after I reduced the train length to 1/2 the original size (down to 45 cars) while the engine sat there in Run 8 did the train finally begin to crawl forward. Once it did get the train rolling, it got out of the red zone faster on account of the shorter consist.
Compare to the CEFX consist that utilizes MSTS friction values. The lone SD40-2 was capable of getting the entire 90 car train moving.
I can understand how the real world train weights, the 'doubles', and helper assignments found in mountainous western US railroading would have been very hard to replicate successfully when using the stickier Davis values. Even in eastern railroading, how many years did Conrail dispatch 110 car coal trains between Altoona and Harrisburg with only 2 older SD40 types for power? I'm sure at some point engine failure would have resulted in only one SD40 bringing the train home to the end terminal.
Sometimes we get a little too involved in the math that we really don't see how our end result fails at actually mirroring real world conditions.
I created two 90 car trains of 286k covered hoppers (about 12,500 US tons) and attached a single SD40-2. The brown BNSF cars are from Seligman 2 route and have Davis values. The identical gray CEFX cars are from Mullan route and have MSTS Friction values. The only editing done to wag files was to reduce the mass for the seligman car to match that of the mullan car.
With a solid train of Davis values, I could not get the train moving with the single SD40-2. Only after I reduced the train length to 1/2 the original size (down to 45 cars) while the engine sat there in Run 8 did the train finally begin to crawl forward. Once it did get the train rolling, it got out of the red zone faster on account of the shorter consist.
Compare to the CEFX consist that utilizes MSTS friction values. The lone SD40-2 was capable of getting the entire 90 car train moving.
I can understand how the real world train weights, the 'doubles', and helper assignments found in mountainous western US railroading would have been very hard to replicate successfully when using the stickier Davis values. Even in eastern railroading, how many years did Conrail dispatch 110 car coal trains between Altoona and Harrisburg with only 2 older SD40 types for power? I'm sure at some point engine failure would have resulted in only one SD40 bringing the train home to the end terminal.
Sometimes we get a little too involved in the math that we really don't see how our end result fails at actually mirroring real world conditions.
...but I'm no physics authority so I'll just sit back and watch them justify why the "advanced" physics are simply not realistic. 
Comment