It may be instructive to know how this kind of thing has to be proved.
When you look at a stretch of line and the service position on it, you have to use the industry revenue and loads prediction system (MOIRA) to do it. The DfT won’t accept anything else. So you plug in your projected service and get the answers. Then you take your existing loads (and the GWR database is really good - for the whole year), do a comparison and adjust as necessary, as MOIRA works on whole industry demand spread averages, not location specific. So, for example, your daily peak at a certain location may be slightly different to actual so you manually adjust how your increased demand will be spread through the day.
In the case of the 802 fleet, going load 9 or 10 into Cornwall in the DA2 award period would have meant two things - you needed the equivalent of 5 extra 5 cars in diagram (6 in the fleet) and you killed the case for the Cornish half hourly in the process. There just wasn’t the loadings to justify both, even with the most optimistic future demand forecasts. And if you can’t prove the loadings, you can’t get the stock authorised.
7 cars was looked at because they would have fitted at Long Rock but they wouldn’t have coped with the demand from Plymouth inwards to London so you needed load 9 or 10 on that stretch. But you just couldn’t justify them all the way through to Penzance all the time so instead of ordering 22 x 5 cars, you would have needed 14 x 9. So that was 16 cars more (with stabling issues at the country end) and no Cornish half hourly. You can see why the decision to go 5 car was made.
But, in the medium to long term, you could see a case for longer trains into Cornwall developing and so it was felt to be best to go with the Cornish half hourly now and extend the 802 fleet later. There was also a real fear that if you didn’t get the Cornish half hourly in then, it might never happen and that was something that, in hindsight, GWR and Cornwall got absolutely right.
Most of the time a 5 car is perfectly adequate for Plymouth to Penzance but the operational headache of splitting and joining (which the WR hates) is something that is best avoided if you can. The provision of extra siding space at the Penzance end helps this so you don’t have to split and join most of the starters and finishers (because the space at land locked Long Rock wasn’t there for 9 cars) and the fall off in long distance commuting on the GWML from Parkway and Temple Meads inwards to Paddington has released enough units to allow for constant long train working into Cornwall, a bit earlier than GWR had previously planned for.
Unfortunately, playing fantasy units is out. The DfT have made it plain they do not see GWR getting any more 800 series cars at the moment and, contractually, they want to stick to the current fleets being where they are. I suspect that extending the 5 cars might come into play in the next few years but as part of a wider DfT fleet reorganisation that would involve Cross Country.
Sometimes you have to play a less than ideal game to get a result but, getting back to the original question posed here, if GWR didn’t have 5 cars, Bertie the bus might have had to be used for a longer journey than it had to or alternatively everyone would been crammed into a DMU.