That is a can of worms...

I don't fully understand how that works except to say that the TER of the detached unit is higher than for the semis, which means there is more opportunity to get reductions from elsewhere in the detached - ie by having decent boiler / heating controls etc.

I would say that SAP has its faults however, some of which I understand well and others which I don't. The passivhaus / PHPP lot will tell you that it's very harsh on MVHR although I don't understand this fully - something to do with not appreciating the full heat recovery : power used ratio but I don't know the numbers.

SAP also assumes a steady state of heat transfer in the envelope which is not the reality, but you need very souped up dynamic simulation software to address this, which may not be affordable for most SAP assessors.

Slightly separately, I find it crazy that the Code encourages heat pumps which, as SAP's own results show, emit more CO2 than a gas boiler. Yet because the TER skyrockets when heat pumps are entered, they become the attractive option for the developer. I've heard rumours that heat pumps will not get such an easy ride when the Code / SAP is next updated - can anyone shed any light on this?

ESP-r & programs that use it (like Hot3000) do sub-hourly modeling; this will account for dynamic and non-linear heat loss effects.
http://www.esru.strath.ac.uk/Programs/ESP-r.htm
http://canmetenergy-canmetenergie.nrcan-rncan.gc.ca/eng/software_tools/hot3000.html

Shared walls in semi-detatched buildings can be a significant source of heat loss depending on the construction method. Wood-framed double stud shared walls with an airspace in between is a common construction technique here in Canada and air can escape up into the attic space from the space between the walls.