The simple version is:
At the end of each 4 week accounting period, all delays attributed for that period are totted up.
Each TOC or FOC will have delays caused by them, and delays caused to them. For each of these categories, there is a series of benchmarks for the delay each period. Then for delays caused by the TOC/FOC, this is multiplied by a £/min rate which differs depending on where the total for the period sits in relation to the benchmark.
Similarly, for delays caused to the TOC/FOC (which could be by NR, or other TOCS / FOCs) NR pays the ‘victim’ TOC / FOC a £/min rate depending on where the total sits in relation to the benchmarks.
In almost every case, the rate from TOC to NR is lower than the rate from NR to the TOC. The £/min rates are set out in the track access contracts, and vary considerably by TOC and within service groups within that TOC. Eg the £/min rate for peak Thameslink services on the MML is very significantly higher than the £/min rate for off peak coastway services.
FOCs and some smaller operators have caps on their liabilities, as a single large incident could otherwise wipe out their business.
This really is the simple version, it is a lot more complicated in reality. However it’s all done on a big spreadsheet, and takes a couple of hours to work out each period. The attribution of delay itself is a much larger task, but then that needs doing to get to the root cause of delay so that performance improvements can be targeted.