It has always struck me that there are 3 conflicting points about the length of a train (as opposed to the length of one unit in the train):
- The train must be long enough to carry the number of passengers wanting to use it
- The train must be short enough to fit in all the platforms on the journey
- The DfT, with some input from the TOC, must allow enough units to be leased to allow operation at these levels, including growth over time.
In Northern's case, fitting trains into platforms is key ... and requires flexibility. If they had bought 4 carriage units, then the only possible trains are 4, 8 or 12 carriages - and it can take years to lengthen platforms to accommodate a single step increase. With a mix of 2 and 3 car units, they get the best flexibility ... as you can make a train with any number of carriages from 2 upwards. Growth can come 1 carriage at a time.
Of course, this can only work if you leased enough carriages & units to actually run in multi in this way, or DfT allowed it. Also that there is enough headroom for growth, and for servicing such that short forms don't happen.
Northern don't have that luxury right now. Winding down pacers, without a full complement of new trains, and with refurbishment still in full swing is always going to cause pain on allocations.
So I'm wary of converting every 2 car 195 into a 3 car. It loses flexibility in constructing multi-unit trains ... which you could only regain by gradually adding carriages to 3's to make 4s, and so on. 2's and 3's make for better building blocks in a non-homogeneous railway ... and is the whole point of MUs.
It all falls down when the bean counters prevent a TOC from responding to growing requirements.