You also have to consider than an LNER day coach is likely to do more than one trip per day but the maintenance periodicities are likely to be similar in terms of "operational days" Add in the requirements for multiple locos /drivers per trip and it all adds up.
...and then there's all the additional staffing.
Sleeper stock spend several hours during the day being serviced/prepped/made-up and not earning revenue (unlike your LNER Mk3 which will require much less cleaning/prep and will have this done overnight then back out first thing earning coin).
There's lounge car staff, guard / train manager, a host per every 2 coaches - all this multiplied by the number of portions - I'd expect this is a higher staff per train ratio than a day train and certainly a much higher staff per passenger ratio.
There's the shunters at Edinburgh/Carstairs (and Euston).
Now it's a single franchise, all the "support" operations are not shared around as they were in Scotrail days (IT, sales, marketing, administrative etc.)
Every other night of the week you're putting up London and Scotland crews who are at the "wrong end".
The trains also do one single trip (one set of ticket revenue) per 24 hours - not 2, 3 or 4 diagrams a day as an HST set might do.
There's multiple locos (possibly as many as 12) and drivers used every night (via GB) to run the service.
Could go on... but the Sleeper is clearly far more costly than any other passenger train to run - and much more than the simple capacity comparisons.
In their accounts, the main reason Serco gave for the increased costs / losses were the costs of having to run the old stock longer - that's all the accounts really said, so presumably that's a whole host of costs including continued leasing costs, far higher maintenance costs, costs associated with them failing more regularly (delay repay/compensation etc); lost revenue opportunities from not being able to sell the new product, especially the premium en suite berths... etc.
