I have a very interesting question. Before coming to the question, let me first explain something about the situation. On 11th March, 2015, I came to this website to check the arr/dep trend of the 12420 at NDLS to give me some hope that on 6th May, the day I will have to catch an International flight from DEL T3 at 2100, the train will not be delayed by a significant time. So, as I checked the trend, I also noticed that 12419/12420 share rake with 14211/14212. I don't have any expert knowledge of rake sharing in IR from any resource. By logic only, I'm deriving the knowledge here. Logically, by rake sharing, I mean that only after a train, lets say T1, running from A to B, reaches B, then only the train T2, running from B to either back to A or to some C (which is same case...
more...

in my scenario of interest, i.e. first from A to B to C then from C to B to A and so on in loop) and sharing rake with T1, can depart from B. Now, if I believe I'm right in thinking about rake sharing in IR, then the "late" arrival time of T1, lets suppose TaT1, at B should also be closely linked to the "late" departure time of T2, lets suppose TdT2, at B. I found this hypothesis to be somewhat true after I crosschecked by rigorously comparing the (actual) arr/dep history data (on 11 Mar 2015) for all the train pairs i.e. 14211(T1)(A to B)/12420(T2)(B to C), 12420(T2)(B to C)/12419(T3)(C to B), 12419(T3)(C to B)/14212(T4)(B to A) and 14212(T4)(B to A)/14211(T1)(A to B). Lets say TaTis is 's'cheduled time of 'a'rrival of train Ti and TaTia is 'a'ctual time of 'a'rrival of train Ti. I observed that if (TdTjs-TaTis)=x then (TdTja-TaTia)=x +- 1 hourmax (+- 30 minutes in most cases). So, for pair 12419(T3)(C to B)/14212(T4)(B to A), TaT3 scheduled (TaT3s) is 1455 and TdT4s is 1735 at B (NDLS) on same day i.e. TdT4s-TaT3s=0240 hrs. I believe (logically arriving at the it) it means that Bogies in the rake are not cleaned at the yard since the layover is only 0240 hrs. But there was one case on 3rd March in particular, when my hypothesis failed. Lets discuss that case. On 3 March, TaT3a was 2359 (late by 0904hrs) and TdT4a is 0010 (on 4 march) (late by 0635hrs). So, it means the layover was only 11 minutes (TdT4a-TaT3a=0010 hrs). Now, my question is if we believe the data on the website to actually reflecs the arrival and departure time of the two trains, is something like this even possible in reality considering the logistics of operations in an Indian railway station? Also, if any of you think that there is any (unintentional) flaw in the way I've described the situation, please feel free to ask.
Thanks.