It depends how gung-ho you want to be about building AGR stations. I went with Magnox largely to be conservative, and to match the 500 MWe generator set capacity.
Grain was actually built as 3,300 MW from five units; I'm not sure when it was reduced to two units. In addition, Littlebrook is within the boundaries of Greater London; if a nuclear plant was substituted, a different location would definitely be needed. That totals four twin-unit plants, plus Torness and Heysham 2 - which might be AGRs or SHGWRs.
There were also Inverkip and Peterhead oil-fired stations in Scotland, under the SSEB and NoSHEB respectively. Inverkip had three 660 MWe units, with provision for a fourth, but barely ran between 1976 and 1988 when it was decommissioned. You could conceive of an AGR plant or two taking its place, but they might well wind up being white elephants.
Peterhead had two 660 MWe units; an alternative SGHWR plant at Stake Ness was seriously considered for a time. It had the significant advantage of being located close to the landfall of most of the North Sea pipelines, making it somewhat comparable to the pithead coal fired plants. Stake Ness was probably a hard sell in comparison.
You'd have to allow for the cost of providing alternative generating capacity, whether coal, oil or light-water reactors. A 1977 paper argued that the AGR programme had cost £2.1 billion in then-year pounds more than an equivalent LWR programme. Whether that's capital expenditure, or expected lifetime operating costs, I'm not sure.
In 1980, Torness was estimated to cost £1.1 billion to build - suggesting that the £2.1 billion figure might be capital expenditure. Some very rough estimating suggests that the difference in cost between an AGR plant and an LWR plant is comparable to the cost of one aircraft carrier. And there were seven of them.
